WO2021212760A1 - Method and apparatus for determining identity type of person, and electronic system - Google Patents

Abstract

The present application provides a method and apparatus for determining an identity type of a person, and an electronic system. The method comprises: obtaining an image feature of a target image comprising a target person and an image feature of an image to be compared comprising a person having a known identity type, wherein the image feature at least comprises a human body appearance feature of the person, and the image feature of the image to be compared can reflect the identity type of the person having the known identity type; comparing the image features of the target image and the image to be compared, to obtain a feature similarity; determining an identity similarity between the target person and the person having the known identity type according to the feature similarity; and determining the identity type of the target person according to the identity similarity. According to the method, the identity type of the target person without identity information can be determined, and the accuracy is high.

Description

Translated fromChinese

确定人员身份类型的方法、装置和电子系统Method, device and electronic system for determining type of person's identity

相关申请的交叉引用Cross-references to related applications

本申请要求于2020年04月24日提交中国专利局的申请号为2020103369189、名称为“确定人员身份类型的方法、装置和电子系统”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。This application claims the priority of the Chinese patent application with the application number 2020103369189 and titled "Method, Apparatus and Electronic System for Determining the Type of Person's Identity" submitted to the China Patent Office on April 24, 2020, the entire content of which is incorporated by reference In this application.

技术领域Technical field

本申请涉及图像处理技术领域，尤其是涉及一种确定人员身份类型的方法、装置和电子系统。This application relates to the field of image processing technology, and in particular to a method, device and electronic system for determining the identity of a person.

背景技术Background technique

为了方便管理不同人员信息，可以用多维特征集来记录不同的人员信息，多维特征集中可以包含用户的图像集合和身份信息。然而，只有少量人员的多维特征集中包含有人员的身份信息，大多数人员的身份信息未知。In order to facilitate the management of different personnel information, a multi-dimensional feature set can be used to record different personnel information. The multi-dimensional feature set can contain the user's image collection and identity information. However, only a small number of people’s multidimensional feature sets contain their identity information, and most of them’s identity information is unknown.

发明内容Summary of the invention

有鉴于此，本申请的目的之一在于提供一种确定人员身份类型的方法、装置和电子系统，以准确地确定没有身份信息的目标人员的身份类型。In view of this, one of the objectives of the present application is to provide a method, device and electronic system for determining the identity type of a person, so as to accurately determine the identity type of a target person without identity information.

第一方面，本申请实施例提供了一种确定人员身份类型的方法，方法包括：获取包含有目标人员的目标图像的图像特征，以及包含已知身份类型人员的待比对图像的图像特征；其中，所述目标图像的图像特征至少包括所述人员的人体外形特征；所述待比对图像的图像特征用于表征所述已知身份类型人员的身份类型；比对目标图像和待比对图像的图像特征，得到特征相似度；根据特征相似度，确定目标人员和已知身份类型人员的身份相似度；根据身份相似度，确定目标人员的身份类型。In the first aspect, an embodiment of the present application provides a method for determining the identity type of a person. The method includes: acquiring image features of a target image containing a target person, and image features of an image to be compared containing a person with a known identity type; Wherein, the image feature of the target image includes at least the human body shape feature of the person; the image feature of the image to be compared is used to characterize the identity type of the person with the known identity type; the target image is compared with the person to be compared According to the image characteristics of the image, the feature similarity is obtained; according to the feature similarity, the identity similarity between the target person and the person with a known identity type is determined; according to the identity similarity, the identity type of the target person is determined.

可选地，上述待比对图像包括多张；每张待比对图像中包含一个已知身份类型人员；多张待比对图像中包含的已知身份类型人员的身份类型相同；比对目标图像和待比对图像的图像特征，得到特征相似度的步骤，包括：比较每张目标图像的图像特征与每张待比对图像的图像特征，得到多个特征相似度；根据特征相似度，确定目标人员和已知身份类型人员的身份相似度的步骤，包括：根据多个特征相似度，确定目标人员和已知身份类型人员的身份相似度。Optionally, the above-mentioned image to be compared includes multiple images; each image to be compared includes a person with a known identity type; the identity types of persons with known identities contained in the multiple images to be compared are the same; the comparison target The step of obtaining the feature similarity between the image features of the image and the image to be compared includes: comparing the image feature of each target image with the image feature of each image to be compared to obtain multiple feature similarities; according to the feature similarity, The step of determining the identity similarity between the target person and the person with the known identity type includes: determining the identity similarity between the target person and the person with the known identity type according to the similarity of multiple features.

可选地，上述目标图像包括多张；多张目标图像包含同一目标人员；根据多个特征相似度，确定目标人员和已知身份类型人员的身份相似度的步骤，包括：针对每张待比对图像，获取该待比对图像相对于多张目标图像的多个特征相似度中的相似度最大值；将多张待比对图像对应的多个相似度最大值的平均值或最大值，确定为身份相似度。Optionally, the above-mentioned target image includes multiple images; the multiple target images contain the same target person; the step of determining the identity similarity between the target person and the person with a known identity type according to the multiple feature similarities includes: For an image, obtain the maximum similarity among multiple feature similarities of the image to be compared with respect to multiple target images; the average or maximum value of the multiple maximum similarities corresponding to the multiple images to be compared is obtained, Determined as identity similarity.

可选地，上述目标图像包括多张；多张目标图像包含同一目标人员；根据多个特征相似度，确定目标人员和已知身份类型人员的身份相似度的步骤，包括：针对每张目标图像，针对每张所述待比对图像，获取该待比对图像相对于多张所述目标图像的多个特征相似度中的相似度最大值；将多张目标图像对应的多个相似度平均值的最大值，确定为身份相似度。Optionally, the above-mentioned target image includes multiple images; the multiple target images contain the same target person; the step of determining the identity similarity between the target person and the person with a known identity type according to the multiple feature similarity includes: for each target image For each of the images to be compared, obtain the maximum similarity among the feature similarities of the image to be compared with respect to the multiple target images; average the multiple similarities corresponding to the multiple target images The maximum value of the value is determined as the identity similarity.

可选地，所述目标图像包括多张；多张所述目标图像包含同一目标人员；所述比对所述目标图像和所述待比对图像的图像特征，得到特征相似度的步骤，包括：针对每张所述目标图像，比对该目标图像与所述待比对图像的图像特征，得到该目标图像对应的特征相似度；所述根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：将多张所述目标图像对应的特征相似度中的相似度最大值，确定为所述目标人员和所述已知身份类型人员的身份相似度。Optionally, the target image includes multiple images; the multiple target images include the same target person; the step of comparing image features of the target image and the image to be compared to obtain feature similarity includes : For each of the target images, compare the image features of the target image and the image to be compared to obtain the feature similarity corresponding to the target image; the target person and the target person are determined according to the feature similarity. The step of the identity similarity of the person with the known identity type includes: determining the maximum similarity among the feature similarities corresponding to the plurality of target images as the difference between the target person and the person with the known identity type Identity similarity.

可选地，上述包含有目标人员的目标图像预先存储在目标人员的第一人员多维特征集中；如果目标图像包括多张，在多张目标图像中，包含有同一个人员的目标图像存储在同一个第一人员多维特征集中。Optionally, the above-mentioned target image containing the target person is pre-stored in the first person multi-dimensional feature set of the target person; if the target image includes multiple images, among the multiple target images, the target image containing the same person is stored in the same A first-person multi-dimensional feature set.

可选地，上述包含已知身份类型人员的待比对图像预先存储在已知身份类型人员的第二人员多维特征集中；如果待比对图像包括多张，多张待比对图像中，包含有同一个人员的待比对图像存储在同一个第二人员多维特征集中。Optionally, the above-mentioned image to be compared containing persons with known identity types is pre-stored in the second person multi-dimensional feature set of persons with known identity types; if the image to be compared includes multiple images, the multiple images to be compared include The images to be compared for the same person are stored in the same second person multidimensional feature set.

可选地，所述待比对图像包括多张；多张所述待比对图像存储在至少两个所述第二人员多维特征集中；至少两个所述第二人员多维特征集中的待比对图像所包含的所述已知身份类型人员的身份类型相同；所述目标图像包括多张；多张所述目标图像存储在同一个第一人员多维特征集中；所述比对所述目标图像和所述待比对图像的图像特征，得到特征相似度的步骤，包括：针对每个所述第二人员多维特征集，执行下述操作：针对该第二人员多维特征集中的每张所述待比对图像，比较多张所述目标图像的图像特征与该待比对图像的图像特征，得到多张所述目标图像相对于该待比对图像的多个特征相似度；所述根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：针对每个所述第二人员多维特征集，执行以下操作：针对该第二人员多维特征集中的每张所述待比对图像，获取该待比对图像相对于多张所述目标图像的多个特征相似度中的相似度最大值；将该第二人员多维特征集中多张所述待比对图像对应的多个所述相似度最大值的平均值或最大值，确定为所述目标人员与该第二 人员多维特征集对应的已知身份类型的人员的所述身份相似度；将所述目标人员与多个所述第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为所述身份相似度。Optionally, the image to be compared includes a plurality of images; the plurality of images to be compared are stored in at least two of the second person multi-dimensional feature sets; at least two of the second person multi-dimensional feature sets are to be compared The identity types of the persons with known identity types contained in the images are the same; the target images include multiple images; the multiple target images are stored in the same first person multi-dimensional feature set; the target images are compared The step of obtaining feature similarity with the image features of the image to be compared includes: for each of the second person's multi-dimensional feature set, performing the following operation: for each of the second person's multi-dimensional feature set For the image to be compared, compare the image features of the multiple target images with the image features of the image to be compared to obtain multiple feature similarities of the multiple target images with respect to the image to be compared; According to the feature similarity, the step of determining the identity similarity between the target person and the person with the known identity type includes: for each of the second person multi-dimensional feature sets, the following operations are performed: for the second person's multi-dimensional feature set For each of the images to be compared in the collection, the maximum value of the similarity among the feature similarities of the image to be compared with respect to the multiple target images is obtained; The average value or the maximum value of the multiple maximum similarity values corresponding to the image to be compared is determined as the identity similarity between the target person and the person of the known identity type corresponding to the second person multidimensional feature set; The identity similarity is determined as an average value of a plurality of the identity similarities of the target person and the persons of the known identity type corresponding to the plurality of second person multidimensional feature sets.

可选地，所述待比对图像包括多张；多张所述待比对图像存储在至少两个所述第二人员多维特征集中；至少两个所述第二人员多维特征集中的所述待比对图像所包含的所述已知身份类型人员的身份类型相同；所述目标图像包括多张；多张所述目标图像存储在同一个第一人员多维特征集中；所述比对所述目标图像和所述待比对图像的图像特征，得到特征相似度的步骤，包括：针对每个所述第二人员多维特征集，执行下述操作：针对每张所述目标图像，比较该第二人员多维特征集中的多张所述待比对图像的图像特征与该目标图像的图像特征，得到多张所述待比对图像相对于该目标图像的多个特征相似度；所述根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：针对每个所述第二人员多维特征集，执行以下操作：针对每张所述目标图像，获取该目标图像相对于该第二人员多维特征集中的多张所述待比对图像的多个特征相似度的相似度平均值；将多张所述目标图像对应的多个所述相似度平均值的最大值，确定为所述目标人员与该第二人员多维特征集对应的已知身份类型的人员的所述身份相似度；将所述目标人员与多个所述第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为身份相似度。Optionally, the image to be compared includes a plurality of images; the plurality of images to be compared are stored in at least two of the second person multi-dimensional feature sets; the at least two of the second person multi-dimensional feature sets The identity types of the persons with known identity types contained in the images to be compared are the same; the target images include multiple images; the multiple target images are stored in the same first person multidimensional feature set; The step of obtaining the feature similarity between the image features of the target image and the image to be compared includes: for each of the second person multi-dimensional feature sets, performing the following operation: for each of the target images, comparing the first The image features of the multiple images to be compared and the image features of the target image in the multidimensional feature set of two persons are used to obtain multiple feature similarities of the multiple images to be compared with respect to the target image; According to the feature similarity, the step of determining the identity similarity between the target person and the person with the known identity type includes: performing the following operations for each of the second person multi-dimensional feature sets: for each target image , Acquiring the average similarity of the multiple feature similarities of the target image with respect to the multiple images to be compared in the second person multidimensional feature set; and comparing the multiple similarities corresponding to the multiple target images The maximum value of the average value is determined as the identity similarity between the target person and the person of the known identity type corresponding to the second person multi-dimensional feature set; the target person and the multiple second person multi-dimensional features are determined The average value of the multiple identity similarities of persons with known identity types corresponding to the set is determined as the identity similarity.

可选地，上述根据身份相似度，确定目标人员的身份类型的步骤，包括：如果身份相似度高于预设的相似度阈值，确定目标人员的身份类型为已知身份类型。Optionally, the above step of determining the identity type of the target person based on the identity similarity includes: if the identity similarity is higher than a preset similarity threshold, determining that the identity type of the target person is a known identity type.

可选地，所述第二人员多维特征集包括多个多维特征集组；同一所述多维特征集组内的第二人员多维特征集的身份类型相同；不同所述多维特征集组的第二人员多维特征集的身份类型不同；所述目标图像包括多张；多张所述目标图像存储在同一个第一人员多维特征集中；所述比对所述目标图像和所述待比对图像的图像特征，得到特征相似度的步骤，包括：针对每组所述多维特征集组中的每个第二人员多维特征集，针对该第二人员多维特征集中的每张所述待比对图像，比较所述第一人员多维特征集中的多张所述目标图像的图像特征与该待比对图像的图像特征，得到所述第一人员多维特征集相对于该多维特征集组的多个特征相似度；所述根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：针对每个所述多维特征集组中的每个所述第二人员多维特征集，执行以下操作：针对该第二人员多维特征集中的每张所述待比对图像，获取该待比对图像相对于多张所述目标图像的多个特征相似度中的相似度最大值；将多张所述待比对图像对应的多个所述相似度最大值的平均值或最大值，确定为所述目标人员与该第二人员多维特征集对应的已知身份类型的人员的所述身份相似度；将所述目标人员与同一多维特征集组中的多个第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为目标人员与该多维特征集组的身份相似度。Optionally, the second person's multi-dimensional feature set includes a plurality of multi-dimensional feature set groups; the identity types of the second person's multi-dimensional feature set in the same multi-dimensional feature set group are the same; the second person in a different multi-dimensional feature set group has the same identity type; The identity types of the multi-dimensional feature set of persons are different; the target image includes multiple images; the multiple target images are stored in the same multi-dimensional feature set of the first person; the comparison between the target image and the image to be compared The step of obtaining the feature similarity of the image features includes: for each second person multi-dimensional feature set in each group of the multi-dimensional feature set group, for each of the to-be-compared images in the second person multi-dimensional feature set, Compare the image features of the multiple target images in the multi-dimensional feature set of the first person with the image features of the image to be compared to obtain that the multi-dimensional feature set of the first person is similar to the multiple features of the multi-dimensional feature set group Degree; the step of determining the identity similarity between the target person and the person with the known identity type according to the feature similarity includes: for each second in each of the multi-dimensional feature set groups Person multi-dimensional feature set, perform the following operations: for each of the images to be compared in the second person multi-dimensional feature set, obtain the similarity among the feature similarities of the image to be compared with respect to the multiple target images Degree maximum value; determining the average value or maximum value of the multiple similarity maximum values corresponding to the multiple images to be compared as the known identity type corresponding to the multidimensional feature set of the target person and the second person The identity similarity of the person; the average value of the multiple identity similarities of the target person and the multiple second person multi-dimensional feature sets in the same multi-dimensional feature set group corresponding to the known identity type persons, Determine the identity similarity between the target person and the multi-dimensional feature set group.

可选地，所述第二人员多维特征集包括多个多维特征集组；同一所述多维特征集组内的第二人员多维特征集的身份类型相同；不同所述多维特征集组的第二人员多维特征集的身份类型不同；所述目标图像包括多张；多张所述目标图像存储在同一个第一人员多维特征集中；所述比对所述目标图像和所述待比对图像的图像特征，得到特征相似度的步骤，包括：针对每组多维特征集组中的每个第二人员多维特征集，执行下述操作：针对每张所述目标图像，比较该第二人员多维特征集中的多张所述待比对图像的图像特征与该目标图像的图像特征，得到多张所述待比对图像相对于该目标图像的多个特征相似度；所述根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：针对每个所述多维特征集组中的每个所述第二人员多维特征集，执行以下操作：针对每张所述目标图像，获取该目标图像相对于该第二人员多维特征集中的多张所述待比对图像的多个特征相似度的相似度平均值；将多张所述目标图像对应的多个所述相似度平均值的最大值，确定为所述目标人员与该第二人员多维特征集对应的已知身份类型的人员的所述身份相似度；将所述目标人员与同一所述多维特征集组中的多个第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为目标人员与该多维特征集组的身份相似度。Optionally, the second person's multi-dimensional feature set includes a plurality of multi-dimensional feature set groups; the identity types of the second person's multi-dimensional feature set in the same multi-dimensional feature set group are the same; the second person in a different multi-dimensional feature set group has the same identity type; The identity types of the multi-dimensional feature set of persons are different; the target image includes multiple images; the multiple target images are stored in the same multi-dimensional feature set of the first person; the comparison between the target image and the image to be compared Image features, the step of obtaining feature similarity includes: for each second person's multi-dimensional feature set in each group of multi-dimensional feature set groups, performing the following operation: For each of the target images, compare the second person's multi-dimensional feature set The image features of the multiple images to be compared and the image features of the target image are collected to obtain multiple feature similarities of the multiple images to be compared with respect to the target image; said according to the feature similarity , The step of determining the identity similarity between the target person and the person with the known identity type includes: performing the following operations for each of the second person multi-dimensional feature sets in each of the multi-dimensional feature set groups: For each of the target images, obtain the average similarity of the similarity of the target image with respect to the plurality of feature similarities of the plurality of images to be compared in the second person's multi-dimensional feature set; The maximum value of the average value of the multiple similarities is determined as the identity similarity between the target person and the person of the known identity type corresponding to the second person's multi-dimensional feature set; The average value of the identity similarities of the persons with known identity types corresponding to the multiple second person multi-dimensional feature sets in the multi-dimensional feature set group is determined as the identity similarity between the target person and the multi-dimensional feature set group.

可选地，上述根据身份相似度，确定目标人员的身份类型的步骤，包括：从多组所述多维特征集组对应的身份相似度中，选择最高的身份相似度，将所述最高的身份相似度对应的多维特征集组的身份类型，确定为所述目标人员的身份类型；或者，针对每组所述多维特征集组，判断该多维特征集组对应的身份相似度是否高于该多维特征集组对应的相似度阈值；如果所述多维特征集组对应的身份相似度高于所述多维特征集组对应的相似度阈值，将该多维特征集组对应的身份类型，确定为所述目标人员的身份类型。Optionally, the above step of determining the identity type of the target person based on the identity similarity includes: selecting the highest identity similarity from a plurality of sets of the identity similarities corresponding to the multi-dimensional feature set groups, and setting the highest identity The identity type of the multi-dimensional feature set group corresponding to the similarity is determined as the identity type of the target person; or, for each group of the multi-dimensional feature set group, it is determined whether the identity similarity corresponding to the multi-dimensional feature set group is higher than that of the multi-dimensional feature set group. The similarity threshold corresponding to the feature set group; if the identity similarity corresponding to the multi-dimensional feature set group is higher than the similarity threshold corresponding to the multi-dimensional feature set group, the identity type corresponding to the multi-dimensional feature set group is determined as the The identity type of the target person.

可选地，上述第一人员多维特征集中的目标图像设置有采集时间；第一人员多维特征集中包括多张目标图像；根据身份相似度，确定目标人员的身份类型的步骤之后，方法还包括：根据每张目标图像对应的特征相似度，以及目标图像的采集时间，确定目标人员的工作时间。Optionally, the target image in the first person multi-dimensional feature set is set with a collection time; the first person multi-dimensional feature set includes multiple target images; after the step of determining the identity type of the target person according to the identity similarity, the method further includes: According to the feature similarity corresponding to each target image and the acquisition time of the target image, the working time of the target person is determined.

可选地，上述根据每张目标图像对应的特征相似度，以及目标图像的采集时间，确定目标人员的工作时间的步骤，包括：针对每张所述目标图像，判断该目标图像的特征相似度是否高于预设的相似度阈 值；如果所述目标图像的特征相似度高于所述预设的相似度阈值，确定该目标图像对应的采集时间属于所述目标人员的工作时间；将属于所述目标人员的工作时间的采集时间所组成的时间段，确定为所述目标人员的工作时间。Optionally, the step of determining the working time of the target person according to the feature similarity corresponding to each target image and the collection time of the target image includes: determining the feature similarity of the target image for each target image Whether it is higher than the preset similarity threshold; if the feature similarity of the target image is higher than the preset similarity threshold, it is determined that the acquisition time corresponding to the target image belongs to the working time of the target person; The time period composed of the collection time of the working time of the target person is determined as the working time of the target person.

可选地，上述根据每张目标图像对应的特征相似度，以及目标图像的采集时间，确定目标人员的工作时间的步骤之后，方法还包括：获取预设区域范围内的目标人员的工作时间；根据目标人员的工作时间，确定预设区域范围内，在指定时间点处于工作状态的目标人员的数量。Optionally, after the step of determining the working time of the target person according to the feature similarity corresponding to each target image and the acquisition time of the target image, the method further includes: acquiring the working time of the target person within the preset area; According to the working hours of the target personnel, determine the number of target personnel in the working state at the specified time point within the preset area.

可选地，上述目标人员包括多个；多个目标人员的第一人员多维特征集属于同一预设区域范围；根据身份相似度，确定目标人员的身份类型的步骤之后，方法还包括：根据每个目标人员的第一人员多维特征集所对应的身份类型，从预设区域范围内的目标人员中筛选得到指定人员。Optionally, the aforementioned target person includes multiple; the first person multi-dimensional feature sets of the multiple target persons belong to the same preset area; after the step of determining the identity type of the target person according to the identity similarity, the method further includes: The identity type corresponding to the first person multi-dimensional feature set of each target person is selected from the target person within the preset area to obtain the designated person.

可选地，所述目标图像为未知身份类型的人员的多维特征集中的图像；所述方法还包括：获取所述未知身份类型的人员的图像特征；获取已知身份类型的人员图像；提取所述已知身份类型的人员的图像的图像特征，将所述已知身份类型的人员的图像特征与所述未知身份类型的人员的图像的图像特征进行比对，获得所述未知身份类型的人员与所述已知身份类型的人员的相似度；如果所述未知身份类型的人员与所述已知身份类型的人员的相似度大于预设阈值，确定所述未知身份类型的人员与所述已知身份类型的人员的身份类型相同。Optionally, the target image is an image in a multi-dimensional feature set of persons with unknown identity types; the method further includes: obtaining image features of persons with unknown identity types; obtaining images of persons with known identity types; extracting all The image features of the image of the person with the known identity type are compared, and the image features of the person with the known identity type are compared with the image features of the image of the person with the unknown identity type to obtain the person with the unknown identity type Similarity with the person with the known identity type; if the similarity between the person with the unknown identity type and the person with the known identity type is greater than a preset threshold, it is determined that the person with the unknown identity type is People who know the identity type have the same identity type.

第二方面，本申请实施例提供了一种确定人员身份类型的装置，装置包括：特征获取模块，配置成获取包含有目标人员的目标图像的图像特征，以及包含已知身份类型人员的待比对图像的图像特征；其中，图像特征至少包括人员的人体外形特征；待比对图像的图像特征能够体现已知身份类型人员的身份类型；特征比对模块，配置成比对目标图像和待比对图像的图像特征，得到特征相似度；身份相似度确定模块，配置成根据特征相似度，确定目标人员和已知身份类型人员的身份相似度；身份类型确定模块，配置成根据身份相似度，确定目标人员的身份类型。In a second aspect, an embodiment of the present application provides a device for determining the identity type of a person. The device includes: a feature acquisition module configured to acquire image features of a target image containing a target person, and a waiting comparison that includes a person with a known identity type. The image features of the image; among them, the image features include at least the human body shape feature; the image feature of the image to be compared can reflect the identity type of the person with a known identity type; the feature comparison module is configured to compare the target image with the target image For the image features of the image, the feature similarity is obtained; the identity similarity determination module is configured to determine the identity similarity between the target person and the person with a known identity type according to the feature similarity; the identity type determination module is configured to be based on the identity similarity, Determine the identity type of the target person.

第三方面，本申请实施例提供了一种电子系统，电子系统包括：处理设备和存储装置；存储装置上存储有计算机程序，计算机程序在被处理设备运行时执行上述确定人员身份类型的方法。In a third aspect, an embodiment of the present application provides an electronic system. The electronic system includes: a processing device and a storage device; the storage device stores a computer program, and the computer program executes the above-mentioned method for determining the identity type of a person when the processed device is running.

第四方面，本申请实施例提供了一种机器可读存储介质，机器可读存储介质上存储有计算机程序，计算机程序被处理设备运行时执行如上述确定人员身份类型的步骤。In a fourth aspect, the embodiments of the present application provide a machine-readable storage medium, and a computer program is stored on the machine-readable storage medium. When the computer program is run by a processing device, the steps for determining the identity type of a person as described above are executed.

本申请实施例提供了确定人员身份类型的方法、装置和电子系统，首先获取包含有目标人员的目标图像的图像特征，以及包含已知身份类型人员的待比对图像的图像特征；然后比对目标图像和待比对图像的图像特征，得到特征相似度；进而根据该特征相似度，确定目标人员和已知身份类型人员的身份相似度，根据该身份相似度，确定目标人员的身份类型。该方式中的图像特征至少包括人员的人体外形特征，且待比对图像的图像特征能够体现已知身份类型人员的身份类型，因而该方式可以推测出外形特征能够体现身份类型的目标人员的身份类型，且准确率较高，从而有利于对这些人员进行管理，提高发现某些特定身份类型人员的效率，实现人员身份的自动化分析。The embodiment of the application provides a method, device and electronic system for determining the identity type of a person. First, the image features of the target image containing the target person and the image features of the image to be compared containing the person with a known identity type are obtained; then the comparison is performed The image features of the target image and the image to be compared are used to obtain the feature similarity; then according to the feature similarity, the identity similarity between the target person and the person with a known identity type is determined, and the identity type of the target person is determined according to the identity similarity. The image features in this method include at least the human body shape characteristics of the person, and the image characteristics of the image to be compared can reflect the identity type of the person with a known identity type, so this method can infer that the shape feature can reflect the identity of the target person of the identity type Type, and the accuracy rate is high, which is conducive to the management of these personnel, improve the efficiency of discovering certain types of personnel with specific identities, and realize the automated analysis of personnel identities.

本申请实施例的其他特征和优点将在随后的说明书中阐述，并且，部分地从说明书中变得显而易见，或者通过本申请实施例而了解。本申请的目的和其他优点在说明书、权利要求书以及附图中所特别指出的结构来实现和获得。Other features and advantages of the embodiments of the present application will be described in the following description, and partly become obvious from the description, or can be understood through the embodiments of the present application. The purpose and other advantages of the application are realized and obtained by the structures specifically pointed out in the description, claims and drawings.

为使本申请的上述目的、特征和优点能更明显易懂，下文特举可选实施例，并配合所附附图，作详细说明如下。In order to make the above objectives, features, and advantages of the present application more comprehensible, optional embodiments accompanied with accompanying drawings are described in detail below.

附图说明Description of the drawings

为了更清楚地说明本申请具体实施方式或现有技术中的技术方案，下面将对具体实施方式或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图是本申请的一些实施方式，对于本领域技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。In order to more clearly illustrate the specific embodiments of this application or the technical solutions in the prior art, the following will briefly introduce the drawings that need to be used in the specific embodiments or the description of the prior art. Obviously, the appendix in the following description The drawings are some embodiments of the application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

图1为本申请实施例提供的一种电子系统的结构示意图；FIG. 1 is a schematic structural diagram of an electronic system provided by an embodiment of this application;

图2为本申请实施例提供的一种确定人员身份类型的方法的流程图；FIG. 2 is a flowchart of a method for determining a person's identity type according to an embodiment of the application;

图3为本申请实施例提供的另一种确定人员身份类型的方法的流程图；FIG. 3 is a flowchart of another method for determining a person's identity type provided by an embodiment of the application;

图4为本申请实施例提供的另一种确定人员身份类型的方法的流程图；FIG. 4 is a flowchart of another method for determining a person's identity type provided by an embodiment of the application;

图5为本申请实施例提供的一种确定人员身份类型的装置的结构示意图。FIG. 5 is a schematic structural diagram of an apparatus for determining a person's identity type provided by an embodiment of the application.

具体实施方式Detailed ways

为使本申请实施例的目的、技术方案和优点更加清楚，下面将结合附图对本申请的技术方案进行清楚、完整地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。In order to make the purpose, technical solutions and advantages of the embodiments of this application clearer, the technical solutions of this application will be described clearly and completely in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of this application, not all of them.的实施例。 Example. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative work shall fall within the protection scope of this application.

为了获得没有身份信息的人员的身份类型，本申请实施例提供的一种确定人员身份类型的方法、装置和电子系统，该技术方案可以应用于人员管理、城市管理等多领域中，该技术可采用相应的软件和硬件实现，以下对本申请实施例进行详细介绍。In order to obtain the identity type of a person without identity information, the embodiment of the application provides a method, device and electronic system for determining the identity type of a person. This technical solution can be applied to various fields such as personnel management and city management. Using corresponding software and hardware implementation, the following describes the embodiments of the present application in detail.

首先，参照图1，图1为本申请实施例提供的一种电子系统的结构示意图，图1所示的电子系统100可以为用于实现本申请实施例的确定人员身份类型的方法、装置示例电子系统100。First, referring to FIG. 1, FIG. 1 is a schematic structural diagram of an electronic system provided by an embodiment of this application. Theelectronic system 100 shown in FIG. 1 may be an example of a method and apparatus for determining a person's identity type in this embodiment of the application.Electronic system 100.

如图1所示，，电子系统100可以配置有一个或多个处理设备102、一个或多个存储装置104、输入装置106、输出装置108以及一个或多个图像采集设备110，这些组件可以通过总线系统112和/或其它形式的连接机构(未示出)互连。应当注意，图1所示的电子系统100的组件和结构只是示例性的，而非限制性的，根据需要，所述电子系统也可以具有其他组件和结构。As shown in FIG. 1, theelectronic system 100 can be configured with one or more processing devices 102, one or more storage devices 104, an input device 106, an output device 108, and one or more image acquisition devices 110. These components can be The bus system 112 and/or other forms of connection mechanisms (not shown) are interconnected. It should be noted that the components and structure of theelectronic system 100 shown in FIG. 1 are only exemplary and not restrictive, and the electronic system may also have other components and structures as required.

在一些可能的示例中，处理设备102可以是网关，也可以为智能终端，或者是包含中央处理单元(CPU)或者具有数据处理能力和/或指令执行能力的其它形式的处理单元的设备，可以对所述电子系统100中的其它组件的数据进行处理，还可以控制所述电子系统100中的其它组件以执行期望的功能。In some possible examples, the processing device 102 may be a gateway, or an intelligent terminal, or a device that includes a central processing unit (CPU) or other forms of processing units with data processing capabilities and/or instruction execution capabilities. Processing data of other components in theelectronic system 100 can also control other components in theelectronic system 100 to perform desired functions.

在一些可能的示例中，存储装置104可以包括一个或多个计算机程序产品，所述计算机程序产品可以包括各种形式的机器可读存储介质，例如易失性存储器和/或非易失性存储器。该易失性存储器例如可以包括随机存取存储器(RAM)和/或高速缓冲存储器(cache)等。所述非易失性存储器例如可以包括只读存储器(ROM)、硬盘、闪存等。在机器可读存储介质上可以存储一个或多个计算机程序指令，处理设备102可以运行所述程序指令，以实现下文所述的本申请实施例中(由处理设备实现)的客户端功能以及/或者其它期望的功能。在机器可读存储介质中还可以存储各种应用程序和各种数据，例如应用程序使用和/或产生的各种数据等。In some possible examples, the storage device 104 may include one or more computer program products, and the computer program products may include various forms of machine-readable storage media, such as volatile memory and/or nonvolatile memory. . The volatile memory may include random access memory (RAM) and/or cache memory (cache), for example. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, and the like. One or more computer program instructions can be stored on the machine-readable storage medium, and the processing device 102 can run the program instructions to implement the client functions in the embodiments of the present application described below (implemented by the processing device) and/ Or other desired functions. The machine-readable storage medium may also store various application programs and various data, such as various data used and/or generated by the application program.

在一些可能的示例中，输入装置106可以是用户用来输入指令的装置，并且可以包括键盘、鼠标、麦克风和触摸屏等中的一个或多个。In some possible examples, the input device 106 may be a device used by the user to input instructions, and may include one or more of a keyboard, a mouse, a microphone, and a touch screen.

在一些可能的示例中，输出装置108可以向外部(例如，用户)输出各种信息(例如，图像或声音)，并且可以包括显示器、扬声器等中的一个或多个。In some possible examples, the output device 108 may output various information (for example, images or sounds) to the outside (for example, a user), and may include one or more of a display, a speaker, and the like.

在一些可能的示例中，图像采集设备110可以配置成采集预览视频帧或图像数据，并且将采集到的预览视频帧或图像数据存储在所述存储装置104中以供其它组件使用。In some possible examples, the image capture device 110 may be configured to capture preview video frames or image data, and store the captured preview video frames or image data in the storage device 104 for use by other components.

示例性地，配置成实现根据本申请实施例的确定人员身份类型的方法、装置和电子系统的示例电子系统中的各器件可以集成设置，也可以分散设置，如将处理设备102、存储装置104、输入装置106和输出装置108集成设置于一体，而将图像采集设备110设置于可以采集到目标图像的指定位置。当上述电子系统中的各器件集成设置时，该电子系统可以被实现为诸如相机、智能手机、平板电脑、计算机、车载终端等智能终端。Exemplarily, each device in the example electronic system configured to implement the method, device, and electronic system for determining the type of person's identity according to the embodiments of the present application can be integrated or distributed, such as the processing device 102 and the storage device 104. , The input device 106 and the output device 108 are integrated in one body, and the image capture device 110 is set at a designated location where the target image can be captured. When the various devices in the above electronic system are integrated and arranged, the electronic system can be implemented as an intelligent terminal such as a camera, a smart phone, a tablet computer, a computer, a vehicle-mounted terminal, and the like.

参见图2，图2为本申请实施例提供一种确定人员身份类型的方法的流程图，该方法可以由本实施例中电子系统100中的处理设备执行；该处理设备可以是具有数据处理能力的任何设备或芯片。该处理设备可以独立对接收到的信息进行处理，也可以与服务器相连，共同对信息进行分析处理，并将处理结果上传至云端。该方法包括如下步骤：Referring to FIG. 2, FIG. 2 is a flowchart of a method for determining a person's identity type according to an embodiment of this application. The method can be executed by a processing device in theelectronic system 100 in this embodiment; the processing device can be capable of data processing. Any device or chip. The processing device can independently process the received information, or it can be connected to a server to analyze and process the information together, and upload the processing results to the cloud. The method includes the following steps:

步骤S202，获取包含有目标人员的目标图像的图像特征，以及包含已知身份类型人员的待比对图像的图像特征；其中，图像特征至少包括人员的人体外形特征；待比对图像的图像特征能够体现已知身份类型人员的身份类型；Step S202: Obtain the image features of the target image containing the target person, and the image features of the image to be compared containing the person with a known identity type; wherein the image features include at least the human body shape feature of the person; the image feature of the image to be compared Can reflect the identity type of persons with known identity types;

在一些可能的示例中上述的目标人员可以理解为本实施例中需要确定身份类型的人员；由于无法获得目标人员的身份信息，可以根据其他已知身份类型的人员的图像特征，推理目标人员的身份类型。身份信息例如是姓名、证件号。在一种可能的实现方式中，包含有目标人员的目标图像的图像特征可以通过预设的特征提取网络得到，同理，包含已知身份类型人员的待比对图像的图像特征也可以通过特征提取网络得到。在一种可能的实现方式中，图像特征通过特征提取网络提取得到，该图像特征可以体现目标人员的多种特征属性，如图像中人员的衣着、身材、发型、脸型等外形特征，还可以体现该人员所处的环境特征，该人员周围的环境、物品、建筑等。In some possible examples, the above-mentioned target person can be understood as the person whose identity type needs to be determined in this embodiment; since the identity information of the target person cannot be obtained, the target person’s identity can be inferred based on the image characteristics of other persons with known identity types. Identity type. The identity information is, for example, a name and a certificate number. In a possible implementation, the image features of the target image containing the target person can be obtained through a preset feature extraction network. Similarly, the image feature of the image to be compared containing the person with a known identity type can also be obtained through the feature Extract the network to get. In a possible implementation, the image features are extracted through a feature extraction network. The image features can reflect a variety of feature attributes of the target person, such as the person's clothing, body, hairstyle, face and other appearance features in the image, and can also reflect The characteristics of the environment where the person is located, the environment, objects, buildings, etc. around the person.

在一些可能的示例中图像特征至少可以包括人员的人体外形特征；在一种可能的实现方式中，人体外形特征可以为与人员的身份类型或所属行业相关联的特征，如衣着、帽饰、鞋子或人员佩戴的物品等特征；In some possible examples, the image features can include at least a person’s body shape feature; in a possible implementation, the body shape feature can be a feature associated with the person’s identity type or industry, such as clothing, hats, Features such as shoes or items worn by personnel;

在一些可能的示例中，上述图像特征还可以包括人员所处环境的环境特征，该环境特征可以是与人员的身份类型或所属行业相关联的特征，例如，人员周边的交通工具、建筑、环境类型等；环境类型可以为社区、公路、田野等环境类型。In some possible examples, the above-mentioned image features may also include the environmental characteristics of the environment in which the person is located. The environmental characteristics may be characteristics associated with the person’s identity type or industry, for example, vehicles, buildings, and environment around the person. Types, etc.; environmental types can be community, highway, field and other environmental types.

在一些可能的示例中，待比对图像的图像特征可以用来体现已知身份类型人员的身份类型；例如，该待比对图像中的人员通常穿有能够体现已知身份类型的制服，或者待比对图像中的人员处在能够体现已知身份类型的环境中，此时待比对图像的图像特征可以体现已知身份类型人员的身份类型。In some possible examples, the image features of the image to be compared can be used to reflect the identity type of a person with a known identity type; for example, the person in the image to be compared usually wears a uniform that can reflect a known identity type, or The person in the image to be compared is in an environment that can reflect the known identity type. At this time, the image characteristics of the image to be compared can reflect the identity type of the person with the known identity type.

在一些可能的示例中，上述包含有目标人员的目标图可以为全景图像或者是该全景图的截图；前景图像可以由摄像设备拍摄得到，截图可以是通过行人检测算法对全景图像进行行人检测，并将全景图中包含行人的图像区域进行截图得到的图像。In some possible examples, the above-mentioned target image containing the target person may be a panoramic image or a screenshot of the panoramic image; the foreground image may be captured by a camera device, and the screenshot may be a pedestrian detection algorithm on the panoramic image. And the image obtained by taking a screenshot of the image area containing the pedestrian in the panorama.

在一些可能的实现方式中，当目标图像为全景图像，目标图像的图像特征中可以体现目标人员较为全面的属性特征，如目标人员的外形特征和环境特征；当目标图像为截图图像，该目标图像可以包含目标人员一部分属性特征，如包含目标人员的外形特征，以及很少一部分环境特征，甚至不包含环境特征。In some possible implementations, when the target image is a panoramic image, the image features of the target image can reflect the more comprehensive attributes of the target person, such as the shape and environment characteristics of the target person; when the target image is a screenshot image, the target The image can contain part of the attributes of the target person, such as the appearance of the target person, and a small part of the environmental characteristics, or even no environmental characteristics.

步骤S204，比对目标图像和待比对图像的图像特征，得到特征相似度；Step S204, comparing the image features of the target image and the image to be compared to obtain the feature similarity;

在一些可能的实现方式中，可以通过计算图像特征距离的方式，比对目标图像和待比对图像的图像特征；在本申请实施例中，上述的图像特征距离可以理解为特征相似度，在一些可能的实现方式中，可以通过欧氏距离、曼哈顿距离、切比雪夫距离等距离计算方式，计算目标图像和待比对图像的图像特征距离。In some possible implementations, the image feature distance of the target image can be compared with the image feature of the image to be compared by calculating the image feature distance; in the embodiment of this application, the above image feature distance can be understood as the feature similarity. In some possible implementation manners, distance calculation methods such as Euclidean distance, Manhattan distance, and Chebyshev distance may be used to calculate the image feature distance between the target image and the image to be compared.

可以理解的是，目标图像和待比对图像的图像特征距离越近，目标图像和待比对图像的特征相似度越高；目标图像和待比对图像的图像特征距离越远，目标图像和待比对图像的特征相似度越低。It is understandable that the closer the image feature distance between the target image and the image to be compared, the higher the feature similarity between the target image and the image to be compared; the farther the image feature distance between the target image and the image to be compared, the longer the target image and The feature similarity of the image to be compared is lower.

步骤S206，根据特征相似度，确定目标人员和已知身份类型人员的身份相似度；Step S206: Determine the identity similarity between the target person and the person with a known identity type according to the feature similarity;

可以理解，目标图像和待比对图像的图像特征的特征相似度越高，目标人员和已知身份类型人员的身份相似度也就越高。It can be understood that the higher the feature similarity of the image features between the target image and the image to be compared, the higher the identity similarity between the target person and the person with a known identity type.

步骤S208，根据身份相似度，确定目标人员的身份类型。Step S208: Determine the identity type of the target person according to the identity similarity.

在一些可能的示例中，例如，如果身份相似度较高，则可以确定目标人员的身份类型与待比对图像中包含的人员的身份类型相同，即目标人员的身份类型为上述已知身份类型。如果身份相似度较低，则可以确定目标人员的身份类型与待比对图像中包含的人员的身份类型不同，即目标人员的身份类型不是上述已知身份类型。In some possible examples, for example, if the identity similarity is high, it can be determined that the identity type of the target person is the same as the identity type of the person contained in the image to be compared, that is, the identity type of the target person is the aforementioned known identity type . If the identity similarity is low, it can be determined that the identity type of the target person is different from the identity type of the person included in the image to be compared, that is, the identity type of the target person is not the aforementioned known identity type.

在一些可能的示例中，，上述已知身份类型可以为多种；在一种可能的实现方式中，为了获得身份相似度，可以将目标图像的图像特征逐一与包含每种已知身份类型人员的待比对图像的图像特征进行比对，得到目标人员与每种已知身份类型人员的身份相似度，选取身份相似度较高的已知身份类型，确定为目标人员的身份类型。在一些可能的示例中，该身份类型可以为快递外卖人员、环卫人员、保安人员、武警、医院工作人员、警察等身份类型。这些身份类型的人员的外形、环境等通常具有较为统一的特征，通过比对图像特征，可以推理得到没有身份信息的人员的身份类型。In some possible examples, the above-mentioned known identity types can be multiple; in a possible implementation, in order to obtain identity similarity, the image features of the target image can be combined with each of the known identity types. The image features of the images to be compared are compared to obtain the identity similarity between the target person and each known identity type, and the known identity type with higher identity similarity is selected to determine the identity type of the target person. In some possible examples, the identity type may be an identity type such as express delivery personnel, environmental sanitation personnel, security personnel, armed police, hospital staff, and police. The appearance, environment, etc. of persons with these identity types usually have relatively uniform characteristics. By comparing image features, the identity type of persons without identity information can be inferred.

需要说明的是，本实施例中确定人员身份类型的方法可以得到目标人员的身份类型，无法直接得到该目标人员的身份信息；但是，得到目标人员的身份类型，也有助于缩小对目标人员调查的范围，便于对目标人员进行管理。It should be noted that the method for determining the identity type of a person in this embodiment can obtain the identity type of the target person, but cannot directly obtain the identity information of the target person; however, obtaining the identity type of the target person also helps to narrow down the investigation of the target person. It is convenient to manage the target personnel.

上述确定人员身份类型的方法，获取包含有目标人员的目标图像的图像特征，以及包含已知身份类型人员的待比对图像的图像特征；然后比对目标图像和待比对图像的图像特征，得到特征相似度；进而根据该特征相似度，确定目标人员和已知身份类型人员的身份相似度，根据该身份相似度，确定目标人员的身份类型。该方式中的图像特征至少包括人员的人体外形特征，且待比对图像的图像特征能够体现已知身份类型人员的身份类型，因而该方式可以推测出外形特征能够体现身份类型的目标人员的身份类型，且准确率较高，从而有利于对这些人员进行管理，提高发现某些特定身份类型人员的效率，实现人员身份的自动化分析。The above method for determining the identity type of a person obtains the image characteristics of the target image containing the target person and the image characteristics of the image to be compared that includes the person with a known identity type; then compares the image characteristics of the target image and the image to be compared, Obtain the feature similarity; then according to the feature similarity, determine the identity similarity between the target person and the person with a known identity type, and determine the identity type of the target person according to the identity similarity. The image features in this method include at least the human body shape characteristics of the person, and the image characteristics of the image to be compared can reflect the identity type of the person with a known identity type, so this method can infer that the shape feature can reflect the identity of the target person of the identity type Type, and the accuracy rate is high, which is conducive to the management of these personnel, improve the efficiency of discovering certain types of personnel with specific identities, and realize the automated analysis of personnel identities.

本实施例提供另一种确定人员身份类型的方法，在本实施例提供的确定人员身份类型的方法中，为了提高确定目标人员的身份类型的准确度，待比对图像可以包括多张；每张待比对图像中可以包含一个已知身份类型人员；多张待比对图像中包含的已知身份类型人员的身份类型相同。This embodiment provides another method for determining the identity type of a person. In the method for determining the identity type of a person provided in this embodiment, in order to improve the accuracy of determining the identity type of the target person, the images to be compared may include multiple images; The images to be compared may contain a person with a known identity type; the identity types of persons with known identities contained in multiple images to be compared are the same.

可以理解的是，在比对目标图像和待比对图像的图像特征的实现方式中，可以比较每张目标图像的图像特征与每张待比对图像的图像特征，得到多个特征相似度；进而根据多个特征相似度，确定目标人员和已知身份类型人员的身份相似度。例如，在一种可能的示例中，目标图像可以为一张，而待比对图像可以有多张，将该目标图像的图像特征与每张待比对图像的图像特征进行比对之后，得到每张待比对图像相对于目标图像的特征相似度，即上述多个特征相似度；为了确定最终的身份相似度，可以对多个特征相似度进行后续处理，从而得到目标人员和已知身份类型人员的身份相似度；例如，在一种可能的处理方式中，可以对多个特征相似度进行平均、加权平均等处理，得到身份相似度；在另一种可能的处理方式中，在对多个特征相似度进行平均、加权平均的过程中，可以去掉多个特征相似度的一个或多个最大值，或者去掉多个特征相似度的一个或多个最小值，然后再对剩余的特征相似度进行平均、加权平均等处理。在另一种可能的处理方式中，还可以将多个特征相似度中的最大值或最小值，确定为身份相似度。It is understandable that in the implementation of comparing the image characteristics of the target image and the image to be compared, the image characteristics of each target image can be compared with the image characteristics of each image to be compared to obtain multiple feature similarities; Then, according to the similarity of multiple features, the identity similarity between the target person and the person with known identity type is determined. For example, in a possible example, the target image may be one image, and there may be multiple images to be compared. After comparing the image characteristics of the target image with the image characteristics of each image to be compared, the result is The feature similarity of each image to be compared relative to the target image, that is, the above-mentioned multiple feature similarities; in order to determine the final identity similarity, multiple feature similarities can be followed up to obtain the target person and the known identity The identity similarity of the types of people; for example, in one possible processing method, multiple feature similarities can be averaged, weighted average, etc., to obtain identity similarity; in another possible processing method, In the process of averaging and weighted average of multiple feature similarities, one or more maximum values of multiple feature similarities can be removed, or one or more minimum values of multiple feature similarities can be removed, and then the remaining features The similarity is processed by average and weighted average. In another possible processing manner, the maximum or minimum value among the multiple feature similarities can also be determined as the identity similarity.

可以理解的是，通过多张待比对图像与目标图像进行图像特征特性比对，可以进一步提高后续推测目标人员的身份类型的准确性和合理性。例如，在一种可能的示例中，待比对图像包括10张，这10张 待比对图像中的人员的身份类型均是保安，而目标图像中的目标人员的身份实际上不是保安；如果目标图像中的目标人员与上述10张中的一张待比对图像进行比对，这张待比对图像中的人员与目标人员佩戴有相似的配饰，导致二者特征相似度较高，这时就很有可能将目标身份的身份类型确定为保安，导致身份类型推测错误。而如果目标图像与10张待比对图像进行比对，则在一种可能的实现方式上，目标图像与大部分的待比对图像的特征相似度较低，此时，就不会将目标人员的身份类型误判为保安。It is understandable that by comparing multiple images to be compared with the target image for image feature characteristics, the accuracy and rationality of subsequent estimation of the target person's identity type can be further improved. For example, in a possible example, the images to be compared include 10 images, and the identity types of the persons in the 10 images to be compared are all security, but the identity of the target person in the target image is not actually a security; if The target person in the target image is compared with one of the 10 images to be compared. The person in the image to be compared and the target person wear similar accessories, which results in a high degree of similarity between the two features. At that time, it is very likely that the identity type of the target identity is determined as security, leading to an error in guessing the identity type. And if the target image is compared with 10 images to be compared, in a possible implementation, the target image and most of the images to be compared have a low similarity in features. At this time, the target image will not be The identity type of the personnel was misjudged as security.

在本实施例一种可能的实现方式中，目标图像可以包括多张，也可以包括一张，当目标图像包括多张时，多张目标图像中可以包含同一目标人员，该目标人员即需要确定身份类型的人员。基于此，本申请实施例给出一种确定身份相似度的实现方式，即针对每张待比对图像，获取每张待比对图像相对于多张目标图像的多个特征相似度中的相似度最大值；将多张待比对图像对应的多个相似度最大值的平均值或最大值，确定为身份相似度。In a possible implementation of this embodiment, the target image may include multiple or one target image. When the target image includes multiple images, the multiple target images may contain the same target person, and the target person needs to be determined Identity type personnel. Based on this, the embodiment of the present application provides an implementation method for determining identity similarity, that is, for each image to be compared, the similarity among multiple feature similarities of each image to be compared with respect to multiple target images is obtained. Degree maximum value; the average or maximum value of the multiple similarity maximum values corresponding to the multiple images to be compared is determined as the identity similarity.

可以理解的是，当目标图像包括多张时，有时并非每张目标图像的图像特征均能用于表征身份类型；例如，在一种可能的示例中，有的目标图像中的目标人员穿着能够体现身份类型的制服，此时该目标图像的图像特征能够用于表征目标图像中目标人员的身份类型；而有的目标图像中的目标人员穿着便装，此时该目标图像的图像特征不能用于表征目标图像中目标人员的身份类型；在各目标图像包含的图像特征对身份类型的表征能力不同的情况下，本申请实施例的还给出一种可能的确定身份相似度的实现方式，即通过将待比对图像与每张目标图像进行特征比对获得多个特征相似度后，从中选取特征相似度的相似度最大值作为当前待比对图像对应的相似度最大值；再将多张待比对图像的相似度最大值中平均值或最大值，确定为身份相似度。It is understandable that when the target image includes multiple images, sometimes not every image feature of the target image can be used to characterize the identity type; for example, in a possible example, the target person in some target image can be dressed Uniforms that reflect the identity type. At this time, the image characteristics of the target image can be used to characterize the identity type of the target person in the target image; while the target person in the target image is dressed in casual clothes, the image characteristics of the target image cannot be used at this time. Characterize the identity type of the target person in the target image; in the case where the image features contained in each target image have different ability to characterize the identity type, the embodiment of the present application also provides a possible implementation method for determining the identity similarity, namely After obtaining multiple feature similarities by comparing the features of the image to be compared with each target image, select the maximum similarity of the feature similarity as the maximum similarity corresponding to the current image to be compared; The average or maximum value among the maximum similarity values of the images to be compared is determined as the identity similarity.

为了方便理解上述确定身份相似度的方式，在一些可能的示例中，假设目标图像包括三张，分别为目标图像A、目标图像B和目标图像C；待比对图像包括四张，分别为待比对图像1、待比对图像2、待比对图像3和待比对图像4。三张目标图像和四张待比对图像的图像特征结果如下述表1所示。In order to facilitate the understanding of the above method of determining identity similarity, in some possible examples, it is assumed that the target image includes three images, namely target image A, target image B, and target image C; The image 1, the image to be compared 2, the image to be compared 3, and the image to be compared 4. The image feature results of the three target images and the four images to be compared are shown in Table 1 below.

表1Table 1

 To目标图像ATarget image A目标图像BTarget image B目标图像CTarget image C待比对图像1Image to be compared 10.80.80.90.90.70.7待比对图像2Image to be compared 20.60.60.70.70.50.5待比对图像3Image to be compared 30.90.90.90.90.80.8待比对图像4Image to be compared 40.50.50.50.50.60.6

上述表1中包括了每张目标图像与每张待比对图像的图像特征的特征相似度，例如，待比对图像1和目标图像A的图像特征的特征相似度为0.8。针对每张待比对图像，获取该待比对图像相对于多张目标图像的多个特征相似度中的相似度最大值；例如，待比对图像1对应的相似度最大值为0.9，待比对图像2对应的相似度最大值为0.7，待比对图像3对应的相似度最大值为0.9，待比对图像4对应的相似度最大值为0.6；然后，将多张待比对图像对应的多个相似度最大值的平均值或最大值，确定为身份相似度。例如，四个待比对图像对应的相似度最大值分别为0.9、0.7、0.9和0.6；其中一种方式中，从这四个相似度最大值中选择最大值，即0.9，作为身份相似度；另一种方式中，将这四个相似度最大值的平均值，即(0.9+0.7+0.9+0.6)/4＝0.775，确定为身份相似度。The foregoing Table 1 includes the feature similarity of the image features of each target image and each image to be compared. For example, the feature similarity of the image features of the image to be compared 1 and the target image A is 0.8. For each image to be compared, obtain the maximum similarity among the feature similarities of the image to be compared with respect to multiple target images; for example, the maximum similarity corresponding to the image 1 to be compared is 0.9, The maximum similarity corresponding to the comparison image 2 is 0.7, the maximum similarity corresponding to the image 3 to be compared is 0.9, and the maximum similarity corresponding to the image 4 to be compared is 0.6; then, multiple images to be compared The average value or the maximum value of the corresponding multiple similarity maximum values is determined as the identity similarity. For example, the maximum similarity values corresponding to the four images to be compared are 0.9, 0.7, 0.9, and 0.6, respectively; in one of the ways, the maximum value of the four similarity maximum values, namely 0.9, is selected as the identity similarity ; In another way, the average value of the four maximum similarities, that is, (0.9+0.7+0.9+0.6)/4=0.775, is determined as the identity similarity.

本申请实施例还提供确定身份相似度的另一种实现方式，例如，当目标图像包括多张时，多张目标图像包含同一目标人员；针对每张目标图像，获取每张目标图像相对于多张待比对图像的多个特征相似度的相似度平均值；将多张目标图像对应的多个相似度平均值的最大值，确定为身份相似度。为了方便理解，请继续以上述表1为例，目标图像A相对于多张待比对图像的多个特征相似度的相似度平均值为(0.8+0.6+0.9+0.5)/4＝0.7；目标图像B相对于多张待比对图像的多个特征相似度的相似度平均值为(0.9+0.7+0.9+0.5)/4＝0.75；目标图像C相对于多张待比对图像的多个特征相似度的相似度平均值为(0.7+0.5+0.8+0.6)/4＝0.65；其中，最大值为0.75，即身份相似度。The embodiment of the present application also provides another implementation method for determining identity similarity. For example, when the target image includes multiple images, the multiple target images include the same target person; for each target image, each target image is obtained relative to the multiple Zhang is to compare the average similarity of multiple feature similarities of the images; the maximum value of the multiple similarity averages corresponding to multiple target images is determined as the identity similarity. To facilitate understanding, please continue to take the above table 1 as an example. The average similarity of the similarity of multiple features of the target image A with respect to the multiple images to be compared is (0.8+0.6+0.9+0.5)/4=0.7; The average similarity of the feature similarity of the target image B with respect to the multiple images to be compared is (0.9+0.7+0.9+0.5)/4=0.75; the target image C is more than that of the multiple images to be compared. The average similarity of the feature similarity is (0.7+0.5+0.8+0.6)/4=0.65; among them, the maximum value is 0.75, which is the identity similarity.

在上述确定身份相似度的另一种实现方式中，针对于每个目标图像，可由多个待比对图像与该目标图像进行特征比对，得到该目标图像的多个特征相似度；将该多个特征相似度的平均值或最大值，确定为该目标图像对应的特征相似度；将多个目标图像对应的特征相似度的最大值，确定为身份相似度。In another implementation manner for determining the identity similarity described above, for each target image, multiple to-be-compared images can be compared with the target image to obtain multiple feature similarities of the target image; The average or maximum value of multiple feature similarities is determined as the feature similarity corresponding to the target image; the maximum value of feature similarities corresponding to the multiple target images is determined as the identity similarity.

本实施例还提供另一种比对图像特征以及确定身份相似度的方式，在一种可能的实现方式中目标图像可以包括多张，该多张目标图像中可以包含同一目标人员，而待比对图像可以包括一张，在这种目标图像可以包括多张，该多张目标图像中可以包含同一目标人员，而待比对图像可以包括一张的情况下，针对每张目标图像，比对该目标图像与待比对图像的图像特征，得到该目标图像对应的特征相似度；将多张目标图像对应的特征相似度中的相似度最大值，确定为目标人员和已知身份类型人员的身份相似度。为了方便理解上述比对图像特征以及确定身份相似度的方式，继续以上述表1为例，在一种可能的示例 中，假设仅存在待比对图像1，三张目标图像与待比对图像1的图像特征进行比对后，得到三个特征相似度，分别为0.8、0.9和0.7，其中的相似度最大值为0.9，即为目标人员和已知身份类型人员的身份相似度。This embodiment also provides another way to compare image features and determine identity similarity. In a possible implementation manner, the target image may include multiple images, and the multiple target images may contain the same target person, and the target image is to be compared. The image may include one image. In this target image, there may be multiple images. The multiple target images may contain the same target person, and the image to be compared may include one image. For each target image, compare The image features of the target image and the image to be compared are obtained, and the feature similarity corresponding to the target image is obtained; the maximum similarity among the feature similarities corresponding to multiple target images is determined as the difference between the target person and the person with a known identity type Identity similarity. In order to facilitate the understanding of the above-mentioned comparison image features and the method of determining identity similarity, continue to take the above-mentioned Table 1 as an example. In a possible example, assume that there is only the image to be compared 1, and three target images and the image to be compared After the image features of 1 are compared, three feature similarities are obtained, respectively 0.8, 0.9, and 0.7. The maximum similarity is 0.9, which is the identity similarity between the target person and the person with a known identity type.

在上述比对图像特征以及确定身份相似度的方式中，可以通过多种方式比对目标图像和待比对图像的图像特征，得到特征相似度，进而可以根据特征相似度，确定目标人员和已知身份类型人员的身份相似度，进而确定目标人员的身份类型。该方式可以推测出外形特征能够体现身份类型的目标人员的身份类型，且准确率较高，从而有利于对这些人员进行管理，提高发现某些特定身份类型人员的效率，实现人员身份的自动化分析。In the above method of comparing image features and determining identity similarity, the image features of the target image and the image to be compared can be compared in a variety of ways to obtain the feature similarity, and then the target person and the existing person can be determined based on the feature similarity. Know the identity similarity of the identity type, and then determine the identity type of the target person. This method can infer the identity type of the target person whose appearance characteristics can reflect the identity type, and the accuracy rate is high, which is conducive to the management of these persons, improves the efficiency of discovering certain persons with specific identity types, and realizes the automated analysis of the identity of the personnel .

本实施例还提供另一种确定人员身份类型的方法，本实施例中，在一种可能的实现方式中，包含有目标人员的目标图像可以预先存储在目标人员的第一人员多维特征集中；第一人员多维特征集中可以包含目标人员的图像集合，如果目标图像包括多张，多张目标图像中，包含有同一个人员的目标图像可以存储在同一个第一人员多维特征集中。可以理解的是，目标人员和第一人员多维特征集之间存在一一对应关系；同一个目标人员的目标图像保存在一个第一人员多维特征集中；不同的目标人员的目标图像保存在不同的第一人员多维特征集中。还可以理解的是，同一第一人员多维特征集中的多张目标图像中，可能包含能够体现身份类型的目标图像，也可能不包含能够体现身份类型的目标图像，可能既包含能够体现身份类型的目标图像(例如，该目标图像中，目标人员穿制服)，又包含不能体现身份类型的目标图像(例如，该目标图像中，目标人员未穿制服)。This embodiment also provides another method for determining the identity type of a person. In this embodiment, in a possible implementation manner, the target image containing the target person may be pre-stored in the first person multi-dimensional feature set of the target person; The multi-dimensional feature set of the first person may include a set of images of the target person. If the target image includes multiple images, the target images containing the same person can be stored in the same multi-dimensional feature set of the first person. It is understandable that there is a one-to-one correspondence between the target person and the first person multi-dimensional feature set; the target image of the same target person is stored in a first person multi-dimensional feature set; the target image of different target persons is stored in different The first person has a multi-dimensional feature set. It is also understandable that the multiple target images in the same first person multi-dimensional feature set may include target images that can reflect the identity type, or may not include the target image that can reflect the identity type, and may include both the target image that can reflect the identity type. The target image (for example, in the target image, the target person is wearing a uniform), it also contains a target image that cannot reflect the identity type (for example, in the target image, the target person is not wearing a uniform).

在一些可能的实现方式中，多张目标图像中包含有同一个人员，该多张目标图像可以位于同一个第一人员多维特征集中；在多张待比对图像中，如果每张待比对图像中包含的人员均不同，此时，多张待比对图像就不会位于多维特征集中了。基于此多张目标图像与多张待比对图像进行特征比对的具体实现方式，可参考前述实施例，在此不再赘述。In some possible implementations, multiple target images contain the same person, and the multiple target images can be located in the same first person multi-dimensional feature set; among multiple images to be compared, if each is to be compared The people included in the images are all different. At this time, multiple images to be compared will not be located in the multi-dimensional feature set. For the specific implementation of feature comparison between the multiple target images and multiple to-be-compared images, reference may be made to the foregoing embodiment, which will not be repeated here.

在另一些可能的实现方式中，上述多张待比对图像还可以被划分为多组，每组中的待比对图像中的人员的身份类型相同，不同组中的待比对图像中的人员的身份类型不同；例如，在一些可能的示例中，上述多张待比对图像可以被划分为环卫工人组、保安组、外卖人员组等。基于此，可以将目标图像与每组待比对图像进行比对，得到目标图像与每组待比对图像的身份相似度，将身份相似度最高的组对应的身份类型，确定为目标图像中目标人员的身份类型。In other possible implementations, the multiple images to be compared may also be divided into multiple groups, and the identity types of the persons in the images to be compared in each group are the same, and the images in the images to be compared in different groups have the same identity type. The identity types of the personnel are different; for example, in some possible examples, the multiple images to be compared can be divided into a sanitation worker group, a security group, a takeaway group, and so on. Based on this, the target image can be compared with each group of images to be compared to obtain the identity similarity between the target image and each group of images to be compared, and the identity type corresponding to the group with the highest identity similarity is determined as the target image The identity type of the target person.

在另一些可能的实现方式，包含已知身份类型人员的待比对图像可以预先存储在已知身份类型人员的第二人员多维特征集中；第二人员多维特征集中可以包含已知身份类型人员的图像集合，如果待比对图像有多张，则将包含有同一个人员的待比对图像存储在同一个第二人员多维特征集中。基于此已知身份类型人员与第二人员多维特征集之间存在一一对应关系。同一个已知身份类型人员的待比对图像保存在一个第二人员多维特征集中；不同的已知身份类型人员的待比对图像保存在不同的第二人员多维特征集中。In other possible implementations, the images to be compared containing persons with known identity types can be pre-stored in the second person multi-dimensional feature set of persons with known identity types; the second person multi-dimensional feature set can include persons with known identity types Image collection, if there are multiple images to be compared, the images to be compared containing the same person are stored in the same second person multidimensional feature set. Based on this known identity type, there is a one-to-one correspondence between the multidimensional feature set of the second person and the person. The to-be-compared images of persons with the same known identity type are stored in a second person multi-dimensional feature set; the to-be-compared images of persons with different known identity types are stored in a different second person multi-dimensional feature set.

在本申请实施例中，本实施例采用至少两个第二人员多维特征集，以确定目标人员的身份类型；基于此，在一种可能的实现方式中，待比对图像可以有多张；多张待比对图像可以存储在至少两个第二人员多维特征集中；每个第二人员多维特征集中的待比对图像所包含的已知身份类型人员的身份类型相同；其中，每个第二人员多维特征集对应一个已知身份类型的人员，不同的第二人员多维特征集对应的已知身份类型的人员不同；因此，该实施例中，可以将包含有相同身份类型人员的待比对图像与目标图像进行特征比对。在另一种实现方式中，上述目标图像可以为多张；多张目标图像可以存储在同一个第一人员多维特征集中。例如，获取两个第二人员多维特征集，每个第二人员多维特征集对应一个人员，这两个人员的身份类型相同，如均为环卫工人，通过这两个第二人员多维特征集中的待比对图像，确定目标人员是否也是环卫工人。为了方便理解上述确定目标人员身份类型的实现过程，请参见图3，图3为本申请实施例提供的另一种确定人员身份类型方法的流程图。In this embodiment of the present application, this embodiment uses at least two second person multi-dimensional feature sets to determine the identity type of the target person; based on this, in a possible implementation manner, there may be multiple images to be compared; Multiple images to be compared can be stored in at least two second person multi-dimensional feature sets; the images to be compared in each second person multi-dimensional feature set contain the same identity types of persons with known identities; The two-person multi-dimensional feature set corresponds to a person with a known identity type, and different second-person multi-dimensional feature sets correspond to different people with a known identity type; therefore, in this embodiment, the waiting list of persons with the same identity type can be compared. Compare the features of the image with the target image. In another implementation manner, the foregoing target images may be multiple; multiple target images may be stored in the same first person multi-dimensional feature set. For example, two second person multi-dimensional feature sets are obtained, and each second person multi-dimensional feature set corresponds to a person. The identity types of the two persons are the same, such as sanitation workers. Through the two second person multi-dimensional feature sets After comparing the images, determine whether the target person is also a sanitation worker. In order to facilitate the understanding of the foregoing implementation process of determining the identity type of the target person, please refer to FIG. 3, which is a flowchart of another method for determining the identity type of a person provided by an embodiment of the application.

如图3所示，本实施例中的确定人员身份类型的方法包括如下步骤：As shown in FIG. 3, the method for determining the identity type of a person in this embodiment includes the following steps:

步骤S302，获取包含有目标人员的目标图像的图像特征，以及包含已知身份类型人员的待比对图像的图像特征；其中，图像特征至少包括人员的人体外形特征；待比对图像的图像特征能够体现已知身份类型人员的身份类型；Step S302: Obtain the image features of the target image containing the target person and the image features of the image to be compared that contains the person with a known identity type; wherein the image features include at least the human body shape feature of the person; the image feature of the image to be compared Can reflect the identity type of persons with known identity types;

步骤S304，针对每个第二人员多维特征集，执行下述操作：针对该第二人员多维特征集中的每个待比对图像，比较多张目标图像的图像特征与该待比对图像的图像特征，得到多张目标图像相对于该待比对图像的多个特征相似度；Step S304: For each second person multi-dimensional feature set, perform the following operation: For each image to be compared in the second person multi-dimensional feature set, compare the image features of multiple target images with the image of the image to be compared Features to obtain multiple feature similarities of multiple target images with respect to the image to be compared;

在一些可能的示例中，每个第二人员多维特征集中可以包括一张或多张待比对图像，通过上述步骤S304可以得到每张待比对图像相对于多张目标图像的多个特征相似度；每张待比对图像对应的特征相似度的个数与目标图像的张数一致，例如，在一种可能的实现方式中，某个第二人员多维特征集中包括三张待比对图像，目标图像包括两张，此时每张待比对图像可以得到两个特征相似度。In some possible examples, each second person multi-dimensional feature set may include one or more images to be compared. Through the above step S304, it can be obtained that each image to be compared has multiple features similar to multiple target images. Degree; the number of feature similarities corresponding to each image to be compared is the same as the number of target images. For example, in a possible implementation, a certain second person multi-dimensional feature set includes three images to be compared Image, the target image includes two images. At this time, two feature similarities can be obtained for each image to be compared.

步骤S306，针对每个第二人员多维特征集，执行以下操作：针对该第二人员多维特征集中的每张待比对图像，获取该待比对图像相对于多张目标图像的多个特征相似度中的相似度最大值；将该第二人员多维特征集中多张待比对图像对应的多个相似度最大值的平均值或最大值，确定为目标人员与该第二人员多维特征集对应的已知身份类型的人员的身份相似度；Step S306: For each second person multi-dimensional feature set, perform the following operations: For each image to be compared in the second person multi-dimensional feature set, obtain multiple features similar to multiple target images of the image to be compared The maximum value of similarity in the second person’s multidimensional feature set; the average or maximum value of the multiple similarity maximums corresponding to the multiple images to be compared in the second person’s multidimensional feature set is determined as the target person corresponding to the second person’s multidimensional feature set The identity similarity of persons with known identity types;

由上述可知，在一种可能的示例中，获得每个待比对图像对应的多个特征相似度，每张待比对图像对应的特征相似度的个数与目标图像的张数一致，从这多个特征相似度中选择的数值最大的特征相似度作为上述相似度最大值，可以理解的是，第二人员多维特征集中的每个待比对图像对应一个相似度最大值；在一种可能的实现方式中，如果第二人员多维特征集中的待比对图像为多张，则可以再将每张待比对图像对应的相似度最大值中的数值最大的相似度确定为上述身份相似度，该身份相似度为目标人员与该第二人员多维特征集对应的已知身份类型的人员的身份相似度；在另一种可能的实现方式中，，可以将每张待比对图像对应的相似度最大值求取平均值，将该平均值确定为目标人员与该第二人员多维特征集对应的已知身份类型的人员的身份相似度。此时，如果第二人员多维特征集为多个，则每个第二人员多维特征集对应一个身份相似度。It can be seen from the above that, in a possible example, multiple feature similarities corresponding to each image to be compared are obtained, and the number of feature similarities corresponding to each image to be compared is the same as the number of target images. The feature similarity with the largest value is selected from the multiple feature similarities as the above-mentioned maximum similarity. It can be understood that each image to be compared in the second person's multi-dimensional feature set corresponds to a similarity maximum; In a possible implementation manner, if there are multiple images to be compared in the second person's multidimensional feature set, then the similarity with the largest value among the maximum values of similarity corresponding to each image to be compared can be determined as the above identity Similarity, the identity similarity is the identity similarity between the target person and the person of the known identity type corresponding to the second person’s multi-dimensional feature set; in another possible implementation, each image to be compared can be compared The corresponding maximum value of similarity is calculated as an average value, and the average value is determined as the identity similarity of the target person and the person of the known identity type corresponding to the second person's multidimensional feature set. At this time, if there are multiple second person multi-dimensional feature sets, each second person multi-dimensional feature set corresponds to an identity similarity.

步骤S308，将目标人员与多个第二人员多维特征集对应的已知身份类型的人员的多个身份相似度的平均值，确定为身份相似度。Step S308: Determine the average of the multiple identity similarities between the target person and the multiple second person multidimensional feature sets corresponding to the known identity types of people as the identity similarity.

步骤S310，如果身份相似度高于预设的相似度阈值，确定目标人员的身份类型为所述已知身份类型。In step S310, if the identity similarity is higher than the preset similarity threshold, it is determined that the identity type of the target person is the known identity type.

为了方便理解上述确定目标人员的身份类型的实现方式，例如，在一种可能的示例中，第二人员多维特征集对应的已知身份类型为环卫工人，目标人员与第二人员多维特征集对应的已知身份类型的身份相似度高于预设的相似度阈值，则可以推测该目标人员的身份类型也是环卫工人；如果目标人员与第二人员多维特征集对应的已知身份类型的身份相似度低于或等于预设的相似度阈值，则可以推测该目标人员的身份类型不是环卫工人；此时，则可以获取其他身份类型的第二人员多维特征集，如保安，继续通过上述步骤判断该目标人员是否是保安，直至推测出目标人员的身份类型。In order to facilitate the understanding of the above-mentioned implementation of determining the identity type of the target person, for example, in a possible example, the known identity type corresponding to the second person multi-dimensional feature set is sanitation worker, and the target person corresponds to the second person multi-dimensional feature set If the identity similarity of the known identity type is higher than the preset similarity threshold, it can be inferred that the identity type of the target person is also a sanitation worker; if the identity type of the target person is similar to the known identity type corresponding to the second person’s multidimensional feature set If the degree is lower than or equal to the preset similarity threshold, it can be inferred that the identity type of the target person is not a sanitation worker; at this time, the second person multi-dimensional feature set of other identity types can be obtained, such as security, and continue to judge through the above steps Whether the target person is a security guard, until the identity type of the target person is inferred.

需要说明的是，如果第二人员多维特征集中的多张待比对图像对应多个身份类型多维特征集，那么在一种可能的实现方式中，可以按照一定的排列规则，将目标图像逐一与每个身份类型的第二人员多维特征集中的待比对图像进行比对；也可以将目标图像同时与每个身份类型的第二人员多维特征集中的待比对图像进行比对。It should be noted that if multiple images to be compared in the second person’s multi-dimensional feature set correspond to multiple identity-type multi-dimensional feature sets, then in a possible implementation manner, the target images can be combined with each other one by one according to a certain arrangement rule. Compare the images to be compared in the multi-dimensional feature set of the second person of each identity type; it is also possible to compare the target image with the images to be compared in the multi-dimensional feature set of the second person of each identity type at the same time.

对应于上述图3的方法，还可以有另外一种可能的实现方式，即在比对目标图像和待比对图像的图像特征时，可以针对每个第二人员多维特征集，执行下述操作：Corresponding to the method in Figure 3, there is another possible implementation, that is, when comparing the image features of the target image and the image to be compared, the following operations can be performed for each second person's multidimensional feature set :

针对每张目标图像，比较该第二人员多维特征集中的多张待比对图像的图像特征与该目标图像的图像特征，得到多张待比对图像相对于该目标图像的多个特征相似度；可以理解的是，将每张目标图像的图像特征与一个第二人员多维特征集中的多个待比对图像的图像特征进行比较，得到每张目标图像对应的多个特征相似度。For each target image, compare the image features of the multiple images to be compared in the second person's multidimensional feature set with the image features of the target image to obtain the multiple feature similarities of the multiple images to be compared with the target image It is understandable that the image features of each target image are compared with the image features of multiple images to be compared in a second person multi-dimensional feature set to obtain multiple feature similarities corresponding to each target image.

然后，针对每个第二人员多维特征集，执行以下操作：针对每张目标图像，获取该目标图像相对于该第二人员多维特征集中的多张待比对图像的多个特征相似度的相似度平均值；将多张目标图像对应的多个相似度平均值的最大值，确定为目标人员与该第二人员多维特征集对应的已知身份类型的人员的身份相似度；最后，将目标人员与多个第二人员多维特征集对应的已知身份类型的人员的多个身份相似度的平均值，确定为身份相似度。Then, for each second person multi-dimensional feature set, perform the following operations: For each target image, obtain the similarity of the multiple feature similarity of the target image with respect to the multiple images to be compared in the second person multi-dimensional feature set Degree average value; the maximum value of the multiple similarity average values corresponding to multiple target images is determined as the identity similarity between the target person and the person of the known identity type corresponding to the second person’s multi-dimensional feature set; finally, the target The average of multiple identity similarities of persons of known identity types corresponding to the person and multiple second person multidimensional feature sets is determined as the identity similarity.

为了方便理解上述确定目标人员的身份类型的实现方式，例如，在一种可能的示例中，目标图像包括三张，某个第二人员多维特征集中包括四张待比对图像；此时，每张目标图像对应四个特征相似度，对这四个特征相似度中计算相似度平均值，则每张目标图像对应一个相似度平均值，三张目标图像总共三个相似度平均值；然后，从这三个相似度平均值中选取最大值，该最大值就是目标人员与该第二人员多维特征集对应的已知身份类型的人员的身份相似度。如果一共有两个第二人员多维特征集，此时得到两个身份相似度，将这两个身份相似度的平均值，确定为目标人员与已知身份类型的人员的最终身份相似度。In order to facilitate the understanding of the foregoing implementation of determining the identity type of the target person, for example, in a possible example, the target image includes three images, and a certain second person multi-dimensional feature set includes four images to be compared; in this case, each Two target images correspond to four feature similarities. Calculate the average similarity of these four feature similarities. Then, each target image corresponds to a similarity average, and the three target images have a total of three similarity averages; then, The maximum value is selected from the average values of the three similarities, and the maximum value is the identity similarity between the target person and the person of the known identity type corresponding to the second person multidimensional feature set. If there are two second person multi-dimensional feature sets, two identity similarities are obtained at this time, and the average of the two identity similarities is determined as the final identity similarity between the target person and the person with a known identity type.

上述方式中，从多维特征集的角度进行图像特征的比对，以及确定身份相似度，该方式可以推测出外形特征能够体现身份类型的目标人员的身份类型，且准确率较高，从而有利于对这些人员进行管理，提高发现某些特定身份类型人员的效率，实现人员身份的自动化分析。In the above method, the image features are compared from the perspective of the multi-dimensional feature set and the identity similarity is determined. This method can infer the identity type of the target person whose identity type can be reflected by the appearance feature, and the accuracy rate is high, which is beneficial to Manage these personnel, improve the efficiency of discovering certain types of personnel with specific identities, and realize automated analysis of personnel identities.

在一些可能的示例中，本实施例还提供另一种确定人员身份类型的方法，即本实施例引入多维特征集组的概念，多维特征集组可以表征为第二人员多维特征集包括多个多维特征集组；同一多维特征集组内的第二人员多维特征集的身份类型相同；不同多维特征集组的第二人员多维特征集的身份类型不同；本申请实施例中的目标图像可以有多张；这多张目标图像可以存储在同一个第一人员多维特征集中。可以理解的是，本实施例中的多维特征集组与身份类型存在一一对应的关系，即一个多维特征集组对应一 种身份类型，基于此，下面给出一种将目标图像的图像特征与多个多维特征集组中的待比对图像的图像特征进行比较，从而确定目标图像中目标人员的身份类型的可能的实现方式In some possible examples, this embodiment also provides another method for determining the identity type of a person. That is, this embodiment introduces the concept of a multi-dimensional feature set group. The multi-dimensional feature set group can be characterized as a second person multi-dimensional feature set including multiple Multi-dimensional feature set group; the identity type of the second person multi-dimensional feature set in the same multi-dimensional feature set group is the same; the identity type of the second person multi-dimensional feature set of different multi-dimensional feature set groups is different; the target image in the embodiment of this application may have Multiple images; these multiple target images can be stored in the same first-person multi-dimensional feature set. It can be understood that there is a one-to-one correspondence between the multi-dimensional feature set group and the identity type in this embodiment, that is, a multi-dimensional feature set group corresponds to one identity type. Based on this, the following gives an image feature of the target image Compare with the image features of the image to be compared in multiple multi-dimensional feature set groups to determine the possible realization of the identity type of the target person in the target image

首先，针对每组多维特征集组中的每个第二人员多维特征集，针对该第二人员多维特征集中的每个待比对图像，将第一人员多维特征集中的多张目标图像的图像特征与该待比对图像的图像特征进行比较，得到第一人员多维特征集相对于该多维特征集组的多个特征相似度；每个待比对图像对应多个特征相似度，该多个相似度的个数与第一人员多维特征集中的多张目标图像的个数一致；在一种可能的示例中，如果第二人员多维特征集包括多张待比对图像，则该第二人员多维特征集对应的特征相似度的数量为每个待比对图像对应的特征相似度的总和；当多维特征集组包括多个第二人员多维特征集，该多维特征集组对应的特征相似度的数量为每个第二人员多维特征集对应的特征相似度的数量的总和。上述第一人员多维特征集相对于该多维特征集组的多个特征相似度的数量，即每个第二人员多维特征集对应的特征相似度的数量的总和。First, for each second person multi-dimensional feature set in each group of multi-dimensional feature sets, for each image to be compared in the second person multi-dimensional feature set, the images of multiple target images in the first person multi-dimensional feature set The feature is compared with the image feature of the image to be compared, and the multiple feature similarities of the first person's multi-dimensional feature set relative to the multi-dimensional feature set group are obtained; each image to be compared corresponds to multiple feature similarities, and the multiple The number of similarities is the same as the number of multiple target images in the first person’s multi-dimensional feature set; in a possible example, if the second person’s multi-dimensional feature set includes multiple images to be compared, the second person’s multi-dimensional feature set includes multiple images to be compared. The number of feature similarities corresponding to the multi-dimensional feature set of persons is the sum of the feature similarities corresponding to each image to be compared; when the multi-dimensional feature set group includes multiple multi-dimensional feature sets of the second person, the features corresponding to the multi-dimensional feature set group are similar The number of degrees is the sum of the number of feature similarities corresponding to each second person's multidimensional feature set. The number of similarities between the multi-dimensional feature set of the first person and the multiple feature sets of the multi-dimensional feature set group, that is, the sum of the number of feature similarities corresponding to each multi-dimensional feature set of the second person.

然后，针对每个多维特征集组中的每个第二人员多维特征集，执行以下操作：针对该第二人员多维特征集中的每张待比对图像，获取该待比对图像的图像特征相对于多张目标图像的图像特征的多个特征相似度中的相似度最大值；将多张待比对图像对应的多个相似度最大值的平均值或最大值，确定为目标人员与该第二人员多维特征集对应的已知身份类型的人员的身份相似度。此时，每个第二人员多维特征集对应一个身份相似度，当多维特征集组包括多个第二人员多维特征集时，该多维特征集组则对应多个身份相似度。Then, for each second person multi-dimensional feature set in each multi-dimensional feature set group, perform the following operations: For each image to be compared in the second person multi-dimensional feature set, obtain the image feature relative of the image to be compared The maximum similarity among the multiple feature similarities of the image features of multiple target images; the average or maximum value of the multiple maximum similarities corresponding to the multiple images to be compared is determined as the target person and the first The identity similarity of persons with known identity types corresponding to the two-person multi-dimensional feature set. At this time, each second person multi-dimensional feature set corresponds to one identity similarity, and when the multi-dimensional feature set group includes multiple second person multi-dimensional feature sets, the multi-dimensional feature set group corresponds to multiple identity similarities.

最后，将目标人员与同一多维特征集组中的多个第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为目标人员与该多维特征集组的身份相似度。由上述可知，同一多维特征集组中的第二人员多维特征集对应的身份类型相同，则目标人员与该多维特征集组的身份相似度，即目标人员的身份类型与该多维特征集组对应的身份类型的身份相似度。Finally, the average value of the multiple identity similarities between the target person and the multiple second person multi-dimensional feature sets in the same multi-dimensional feature set group of persons with known identity types is determined as the target person and the multi-dimensional feature set. The identity similarity of the group. It can be seen from the above that the identity type of the second person in the same multi-dimensional feature set group is the same, the identity similarity between the target person and the multi-dimensional feature set group, that is, the identity type of the target person corresponds to the multi-dimensional feature set group The identity similarity of the identity type.

在一些可能的示例中，基于多维特征集组确定目标人员的身份类型，还可以有另外一种实现方式。In some possible examples, there may be another implementation method for determining the identity type of the target person based on the multidimensional feature set group.

首先，针对每组多维特征集组中的每个第二人员多维特征集，执行下述操作：针对每张目标图像，比较该第二人员多维特征集中的多张待比对图像的图像特征与该目标图像的图像特征，得到多张待比对图像相对于该目标图像的多个特征相似度；可以理解的是，每个目标图像对应的多个特征相似度的个数与所述第二人员多维特征集中的待比对图像的张数一致。First, for each second person's multi-dimensional feature set in each group of multi-dimensional feature set groups, perform the following operation: For each target image, compare the image features of multiple images to be compared in the second person's multi-dimensional feature set with The image features of the target image, the multiple feature similarities of the multiple images to be compared relative to the target image are obtained; it is understandable that the number of multiple feature similarities corresponding to each target image is the same as that of the second The number of images to be compared in the multidimensional feature set of the person is the same.

然后，针对每个多维特征集组中的每个第二人员多维特征集，执行以下操作：针对每张目标图像，获取该目标图像相对于该第二人员多维特征集中的多张待比对图像的多个特征相似度的相似度平均值；将多张目标图像对应的多个相似度平均值的最大值，确定为目标人员与该第二人员多维特征集对应的已知身份类型的人员的身份相似度；Then, for each second person multi-dimensional feature set in each multi-dimensional feature set group, perform the following operations: For each target image, obtain the target image relative to the multiple images to be compared in the second person multi-dimensional feature set The similarity average value of the multiple feature similarities of multiple target images; the maximum value of the multiple similarity average values corresponding to multiple target images is determined as the value of the target person and the person of the known identity type corresponding to the second person’s multi-dimensional feature set Identity similarity

最后，将目标人员与同一多维特征集组中的多个第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为目标人员与该多维特征集组的身份相似度。Finally, the average value of the multiple identity similarities between the target person and the multiple second person multi-dimensional feature sets in the same multi-dimensional feature set group of persons with known identity types is determined as the target person and the multi-dimensional feature set. The identity similarity of the group.

基于上述实施例，下面将介绍在＝根据身份相似度，确定目标人员的身份类型，可以有的多种实现方式：Based on the above-mentioned embodiment, the following will introduce the various implementation methods that can be used to determine the identity type of the target person according to the identity similarity:

在一种可能的方式中，从多组多维特征集组对应的身份相似度中，选择最高的身份相似度，将最高的身份相似度对应的多维特征集组的身份类型，确定为目标人员的身份类型；例如，在一种可能的示例中，共有三组多维特征集组，这三组多维特征集组对应的身份类型分别为环卫工人、保安和外卖人员；这三组多维特征集组对应的身份相似度分别为0.8、0.9和0.6，此时，身份类型为保安的多维特征集组对应的身份相似度最高，则可以确定目标人员的身份类型为保安。In a possible way, the highest identity similarity is selected from the identity similarities corresponding to the multi-dimensional feature set groups, and the identity type of the multi-dimensional feature set corresponding to the highest identity similarity is determined as the target person’s identity. Identity type; for example, in one possible example, there are three sets of multi-dimensional feature set groups, and the identity types corresponding to these three sets of multi-dimensional feature set groups are sanitation workers, security guards, and food delivery personnel; these three sets of multi-dimensional feature set groups correspond to The identity similarity of is 0.8, 0.9 and 0.6. At this time, the multi-dimensional feature set corresponding to the identity type of security has the highest identity similarity, and the identity type of the target person can be determined to be security.

在另一种可能的方式中，针对每组多维特征集组，判断该多维特征集组对应的身份相似度，是否高于该多维特征集组对应的相似度阈值；如果该多维特征集组对应的身份相似度高于该多维特征集组对应的相似度阈值，将该多维特征集组对应的身份类型，确定为目标人员的身份类型。该方式确定出的目标人员的身份类型，可能是一种也可能是多种；继续上述示例，如果每组多维特征集组的相似度阈值均是0.85，则保安对应的多维特征集组的身份相似度均高于该相似度阈值，此时，可以确定目标人员的身份类型为保安。In another possible way, for each multi-dimensional feature set group, determine whether the identity similarity corresponding to the multi-dimensional feature set group is higher than the similarity threshold corresponding to the multi-dimensional feature set group; if the multi-dimensional feature set group corresponds to The identity similarity of is higher than the similarity threshold corresponding to the multi-dimensional feature set group, and the identity type corresponding to the multi-dimensional feature set group is determined as the identity type of the target person. The identity type of the target person determined by this method may be one or multiple; continuing the above example, if the similarity threshold of each multi-dimensional feature set group is 0.85, the identity of the corresponding multi-dimensional feature set group of the security guard The similarity is higher than the similarity threshold. At this time, the identity type of the target person can be determined as security.

在上述可能的实现方式中，从多维特征集组的角度进行图像特征的比对，以及确定身份相似度，每个多维特征集组对应一个身份类型；该方式可以推测出外形特征能够体现身份类型的目标人员的身份类型，且准确率较高，从而有利于对这些人员进行管理，提高发现某些特定身份类型人员的效率，实现人员身份的自动化分析。In the above possible implementation manners, the image features are compared from the perspective of the multi-dimensional feature set group, and the identity similarity is determined. Each multi-dimensional feature set group corresponds to an identity type; this method can infer that the appearance feature can reflect the identity type The identity type of the target personnel is higher and the accuracy rate is high, which is conducive to the management of these personnel, improves the efficiency of discovering certain identity types of personnel, and realizes the automated analysis of personnel identities.

基于上述实施例提供的确定人员身份类型的方法，本实施例提供一种应用实施例，首先，第一人员多维特征集中的目标图像设置有采集时间；第一人员多维特征集中包括多张目标图像；在根据身份相似度，确定目标人员的身份类型的步骤之后，还可以根据每张目标图像对应的特征相似度，以及目标图像的采集时间，确定目标人员的工作时间。Based on the method for determining the identity type of a person provided in the above embodiment, this embodiment provides an application example. First, the target image in the multi-dimensional feature set of the first person is set with a collection time; the multi-dimensional feature set of the first person includes multiple target images After the step of determining the identity type of the target person according to the identity similarity, the working time of the target person can also be determined according to the feature similarity corresponding to each target image and the collection time of the target image.

在一些可能的实现方式中，可以针对每张目标图像，判断该目标图像的特征相似度是否高于预设的相似度阈值；如果该目标图像的特征相似度高于预设的相似度阈值，确定该目标图像对应的采集时间属于目标人员的工作时间；将属于目标人员的工作时间的采集时间所组成的时间段，确定为目标人员的工作时间。In some possible implementations, for each target image, it can be determined whether the feature similarity of the target image is higher than the preset similarity threshold; if the feature similarity of the target image is higher than the preset similarity threshold, It is determined that the collection time corresponding to the target image belongs to the working time of the target person; the time period composed of the collection time belonging to the working time of the target person is determined as the working time of the target person.

例如，在一些可能的示例中，目标人员在工作状态下，会穿着指定的工作制服，或者携带指定的工作工具，或者处在指定的环境中；此时包含有目标人员的目标图像的图像特征中通常能体现出该目标人员在工作状态下的身份类型的相关特征；目标人员在非工作状态下，通常会穿着便装，或者不携带工作工具，或者没有处在指定的环境中，此时包含有目标人员的目标图像的图像特征，就难以体现目标人员在工作状态下的身份类型的相关特征。与目标图像进行比对的待比对图像，可以为该目标人员在工作状态下的图像，也可以为其他人员在工作状态下的图像，这里的其他人员在工作状态下的身份类型与目标人员相同。For example, in some possible examples, when the target person is at work, he will wear a designated work uniform, or carry a designated work tool, or be in a designated environment; this time contains the image characteristics of the target image of the target person It usually reflects the relevant characteristics of the target person’s identity type in the working state; when the target person is not working, he usually wears casual clothes, or does not carry work tools, or is not in a designated environment. At this time, it includes With the image characteristics of the target image of the target person, it is difficult to reflect the relevant characteristics of the target person's identity type in the working state. The image to be compared with the target image can be the image of the target person at work, or the image of other personnel at work. Here, the identity type of the other personnel at work and the target person same.

在一些可能的实现方式中，本申请实施例还可以每隔一定的时间段，采集目标人员的目标图像，同时记录目标图像的采集时间，如果某个目标图像，与待比对图像的特征相似度较低，则可以说明该目标图像中的目标人员没有在工作状态，该目标图像对应的采集时间，不属于目标人员的工作时间。基于此，可以统计在一天之内、一周之内或其他周期内，该目标人员的工作时间，以此对目标人员进行考勤。该考勤方式可以更加准确地统计目标人员的工作时间。In some possible implementation manners, the embodiments of the present application may also collect the target image of the target person at regular intervals, and record the acquisition time of the target image at the same time. If a certain target image has similar features to the image to be compared If the degree is low, it can indicate that the target person in the target image is not in a working state, and the acquisition time corresponding to the target image does not belong to the working time of the target person. Based on this, the working time of the target person can be counted within a day, a week, or other periods, so as to check the attendance of the target person. The attendance method can more accurately count the working time of the target personnel.

在另一种应用实施例中，在根据每张目标图像对应的特征相似度，以及目标图像的采集时间，确定目标人员的工作时间的步骤之后，本申请实施例还可以获取预设区域范围内的目标人员的工作时间；然后根据目标人员的工作时间，确定预设区域范围内，在指定时间点处于工作状态的目标人员的数量。例如，在一些可能的示例中，当目标人员的身份类型为保安时，可以通过上述方式确定在某个地理区域范围内，在具体的时间段处于工作状态的保安的数量，以此来评估该地理区域范围的安全系数。In another application embodiment, after the step of determining the working time of the target person according to the feature similarity corresponding to each target image and the acquisition time of the target image, the embodiment of the present application may also obtain the range of the preset area The working hours of the target personnel; then according to the working hours of the target personnel, determine the number of target personnel in the working state at the specified time within the preset area. For example, in some possible examples, when the identity type of the target person is security, the above method can be used to determine the number of security guards working in a specific period of time within a certain geographic area, so as to evaluate the The safety factor of the geographical area.

在另一种应用实施例中，目标人员可以包括多个；多个目标人员的第一人员多维特征集可以属于同一预设区域范围；在根据身份相似度，确定目标人员的身份类型之后，本申请实施例还可以根据每个目标人员的第一人员多维特征集所对应的身份类型，从预设区域范围内的目标人员中筛选得到指定人员。例如，在一些可能的示例中，多个目标人员经常出现在同一小区范围内，这些目标人员中可能有送外卖人员，也可能是小区居民，此时，通过上述方式确定每个目标人员的身份类型之后，就可以从这些目标人员中筛选得到小区居民。In another application embodiment, the target person may include multiple; the first person multi-dimensional feature set of the multiple target persons may belong to the same preset area; after the identity type of the target person is determined according to the identity similarity, this According to the application embodiment, the designated person can be selected from the target persons in the preset area according to the identity type corresponding to the first person multi-dimensional feature set of each target person. For example, in some possible examples, multiple target persons often appear in the same community. These target persons may be food delivery personnel or residents of the community. In this case, the identity of each target person is determined by the above method After the type, the residents of the community can be screened from these target persons.

本实施例提供另一种确定人员身份类型的方法，该方法还可以用于解决无身份信息的人员多维特征集难以确定人员身份类型的问题，通过已知身份类型的人员的多维特征集，推理出未知身份类型的人员的身份类型，从而实现对无身份信息人员的身份类型的掌握，可以服务于多维研判及数据融合应用。为了方便理解本申请实施例的应用方式，请参见图4，图4为本申请提供的另一种确定人员身份类型方法的流程图，如图4所示，该方法还包括如下步骤：This embodiment provides another method for determining the identity type of a person. This method can also be used to solve the problem that it is difficult to determine the identity type of a person with a multi-dimensional feature set of persons without identity information. Based on the multi-dimensional feature set of a person with a known identity type, inference Identify the identity types of persons with unknown identity types, so as to realize the mastery of the identity types of persons without identity information, which can serve for multi-dimensional research and judgment and data fusion applications. In order to facilitate the understanding of the application mode of the embodiment of this application, please refer to FIG. 4. FIG. 4 is a flowchart of another method for determining a person's identity type provided by this application. As shown in FIG. 4, the method further includes the following steps:

步骤S402，获取未知身份类型的人员的多维特征集；通常一个人员建立一个多维特征集，该多维特征集中包括该人员的图像，以及图像的图像特征；图像具体可以为人员的人体抓拍图、人脸抓拍图等；多维特征集中还可以包括人员的人体特征以及人脸特征等。在一些可能的示例中，上述的未知身份类型的人员可能是外卖人员、环卫人员、保安人员、医院工作人员等身份类型的人员的多维特征集。该未知身份类型的人员的图像可以由摄像头、抓拍机等监控设备获得。可以理解的是，多维特征集中的图像没有任何关于该人员身份类型的标识信息。Step S402: Obtain a multi-dimensional feature set of a person with an unknown identity type; usually a person builds a multi-dimensional feature set that includes the person’s image and the image features of the image; the image can specifically be a person’s human body snapshot, person Face capture images, etc.; the multi-dimensional feature set can also include the human body features and facial features of the person. In some possible examples, the above-mentioned unknown identity type personnel may be a multi-dimensional feature set of identity type personnel such as takeaway personnel, environmental sanitation personnel, security personnel, and hospital staff. The image of the person with the unknown identity type can be obtained by surveillance equipment such as a camera and a capture machine. It is understandable that the images in the multi-dimensional feature set do not have any identification information about the identity type of the person.

步骤S404，获取多张已知身份类型的人员图像；Step S404, acquiring multiple images of persons with known identity types;

在一些可能的示例中，该身份类型具体可以为快递外卖人员，也可以为其他身份类型。如果人员图像中包括多个人员，则需要标识出该身份类型对应的人员。为了提高人员身份类型推理的准确性，已知身份类型的人员图像的数量通常较多，例如，可以为十张。In some possible examples, the identity type may specifically be an express delivery person or other identity types. If the person image includes multiple persons, the person corresponding to the identity type needs to be identified. In order to improve the accuracy of the reasoning of the person's identity type, the number of person images with a known identity type is usually large, for example, it can be ten.

步骤S406，提取已知身份类型的人员图像的图像特征；该图像特征具体可以包括已知身份类型的人员的人体特征；将该图像特征与上述多维特征集中包括的图像的图像特征进行比对。Step S406: Extract the image features of the image of the person with the known identity type; the image feature may specifically include the human body feature of the person with the known identity type; compare the image feature with the image features of the image included in the above-mentioned multi-dimensional feature set.

在一些可能的示例中，上述多维特征集中可以包括M张图像，分别对应M个图像特征；已知身份类型的人员图像为N张，对应N个图像特征。则一共需要N轮比对，每轮比对时，获取一张已知身份类型的人员图像的图像特征，与M张多维特征集中的图像的图像特征进行比对，得到M个特征相似度。例如，假设多维特征集中有2张未知身份类型的人员的图像，已知身份类型的人员图像为10张，那么将每一张已知身份类型的人员图像的图像特征和这2张未知身份类型的人员的图像的图像特征进行比对，可以获得每一张已知身份类型的人员图像对应的2个特征相似度，特征相似度的总和为20个。In some possible examples, the above multi-dimensional feature set may include M images, corresponding to M image features, respectively; there are N images of persons with known identity types, corresponding to N image features. A total of N rounds of comparison are required. In each round of comparison, the image features of an image of a person with a known identity type are obtained and compared with the image features of the images in the M multi-dimensional feature set to obtain M feature similarities. For example, suppose there are 2 images of persons with unknown identity types in the multidimensional feature set, and there are 10 images of persons with known identity types, then the image characteristics of each image of persons with known identity types and these 2 unknown identity types By comparing the image features of the images of the persons of, the similarity of 2 features corresponding to each image of the person of known identity type can be obtained, and the total of the similarity of the features is 20.

步骤S408，在一次比对得到的M个特征相似度中选取最高的一个特征相似度，一共得到N个最高的特征相似度，分别为M1、M2、M3、…、MN；将这N个最高的特征相似度求平均值，得到该多维特征集中人员与已知身份类型的人员的相似度Np。In step S408, the highest feature similarity is selected from the M feature similarities obtained in one comparison, and N highest feature similarities are obtained in total, namely M1, M2, M3, ..., MN; these N highest Calculate the average of the feature similarity of the multi-dimensional feature set, and obtain the similarity Np between the person in the multi-dimensional feature set and the person with a known identity type.

需要说明的是，在一种可能的实现方式中，还可以在一次比对得到的M个特征相似度中选取最高的多个特征相似度，如两个特征相似度，此时一共得到2N个最高的特征相似度。另外，还可以对选取的若干个最高的特征相似度设置权重(例如，N个已知身份类型的人员图像中，有3张图像质量最好，可以将这3张图像对应相似度的权重提高)，每个特征相似度乘以对应的权重后再求和或平均，得到该多维特征集中人员与已知身份类型的人员的相似度Np。It should be noted that, in a possible implementation, the highest multiple feature similarity can be selected from the M feature similarities obtained in one comparison, such as two feature similarities, and a total of 2N features can be obtained at this time. The highest feature similarity. In addition, you can also set weights for several selected highest feature similarities (for example, among the N persons with known identity types, 3 images have the best image quality, and the corresponding similarity weights for these 3 images can be increased. ), the similarity of each feature is multiplied by the corresponding weight and then summed or averaged to obtain the similarity Np between the person in the multi-dimensional feature set and the person with a known identity type.

步骤S410，如果该多维特征集中人员与已知身份类型的人员的相似度Np大于预设阈值R，则可以认为该多维特征集中的人员的身份类型与已知身份类型的人员相同。In step S410, if the similarity Np between the person in the multidimensional feature set and the person with the known identity type is greater than the preset threshold R, it can be considered that the identity type of the person in the multidimensional feature set is the same as the person with the known identity type.

根据上述步骤S402-步骤S410，可以确定该多维特征集中人员是否为快递外卖人员、环卫人员、保安人员、医院工作人员等身份类型。According to the above steps S402 to S410, it can be determined whether the person in the multidimensional feature set is an identity type such as express delivery personnel, environmental sanitation personnel, security personnel, hospital staff, etc.

上述方式中，实现了从已知人员的身份信息推理出未知人员的身份类型信息，从而完成了对未掌握人员的身份属性识别；可以获得海量的人员身份类型信息，大大提高了发现可疑人员的效率，实现人像大数据系统的自动化分析。同时，上述方式解决了传统大数据系统的高度依赖人员身份信息的问题，实现了利用已知推理未知的大数据应用，大大提高了人员多维特征集的可用性。In the above method, the identity type information of the unknown person can be inferred from the identity information of the known person, thereby completing the identification of the identity attribute of the unknown person; a large amount of identity type information of the person can be obtained, which greatly improves the discovery of suspicious persons. Efficiency, to realize automatic analysis of portrait big data system. At the same time, the above method solves the problem of the traditional big data system that is highly dependent on personnel identity information, realizes the use of known reasoning and unknown big data applications, and greatly improves the usability of the multidimensional feature set of personnel.

对应于上述方法实施例，参见图5所示的一种确定人员身份类型的装置的结构示意图，该装置包括：Corresponding to the foregoing method embodiment, refer to the schematic structural diagram of an apparatus for determining a person's identity type shown in FIG. 5, and the apparatus includes:

特征获取模块50，配置成获取包含有目标人员的目标图像的图像特征，以及包含已知身份类型人员的待比对图像的图像特征；其中，图像特征至少包括人员的人体外形特征；待比对图像的图像特征能够体现已知身份类型人员的身份类型；The feature acquisition module 50 is configured to acquire the image features of the target image containing the target person, and the image features of the image to be compared containing the person with a known identity type; wherein the image features include at least the human body shape feature of the person; to be compared The image characteristics of the image can reflect the identity type of a person with a known identity type;

特征比对模块51，配置成比对目标图像和待比对图像的图像特征，得到特征相似度；The feature comparison module 51 is configured to compare the image features of the target image and the image to be compared to obtain feature similarity;

身份相似度确定模块52，配置成根据特征相似度，确定目标人员和已知身份类型人员的身份相似度；The identity similarity determination module 52 is configured to determine the identity similarity between the target person and the person with a known identity type according to the feature similarity;

身份类型确定模块53，配置成根据身份相似度，确定目标人员的身份类型。The identity type determination module 53 is configured to determine the identity type of the target person according to the identity similarity.

上述确定人员身份类型的装置，获取包含有目标人员的目标图像的图像特征，以及包含已知身份类型人员的待比对图像的图像特征；然后比对目标图像和待比对图像的图像特征，得到特征相似度；进而根据该特征相似度，确定目标人员和已知身份类型人员的身份相似度，根据该身份相似度，确定目标人员的身份类型。该方式中的图像特征至少包括人员的人体外形特征，且待比对图像的图像特征能够体现已知身份类型人员的身份类型，因而该方式可以推测出外形特征能够体现身份类型的目标人员的身份类型，且准确率较高，从而有利于对这些人员进行管理，提高发现某些特定身份类型人员的效率，实现人员身份的自动化分析。The above device for determining the identity type of a person obtains the image characteristics of the target image containing the target person and the image characteristics of the image to be compared containing the person with a known identity type; then, the image characteristics of the target image and the image to be compared are compared, Obtain the feature similarity; then according to the feature similarity, determine the identity similarity between the target person and the person with a known identity type, and determine the identity type of the target person according to the identity similarity. The image features in this method include at least the human body shape characteristics of the person, and the image characteristics of the image to be compared can reflect the identity type of the person with a known identity type, so this method can infer that the shape feature can reflect the identity of the target person of the identity type Type, and the accuracy rate is high, which is conducive to the management of these personnel, improve the efficiency of discovering certain types of personnel with specific identities, and realize the automated analysis of personnel identities.

可选地，上述待比对图像包括多张；每张待比对图像中包含一个已知身份类型人员；多张待比对图像中包含的已知身份类型人员的身份类型相同；上述特征比对模块，还可以配置成：比较每张目标图像的图像特征与每张待比对图像的图像特征，得到多个特征相似度；上述身份相似度确定模块，还配置成根据多个特征相似度，确定目标人员和已知身份类型人员的身份相似度。Optionally, the above-mentioned image to be compared includes multiple images; each image to be compared includes a person with a known identity type; the identity types of persons with known identities contained in the multiple images to be compared are the same; The module can also be configured to: compare the image feature of each target image with the image feature of each image to be compared to obtain multiple feature similarities; the aforementioned identity similarity determination module is further configured to be based on multiple feature similarities , To determine the identity similarity between the target person and the person with a known identity type.

可选地，上述目标图像包括多张；多张目标图像包含同一目标人员；上述身份相似度确定模块，还可以配置成：针对每张待比对图像，获取该待比对图像相对于多张目标图像的多个特征相似度中的相似度最大值；将多张待比对图像对应的多个相似度最大值的平均值或最大值，确定为身份相似度。Optionally, the foregoing target images include multiple images; the multiple target images include the same target person; the foregoing identity similarity determination module may also be configured to: for each image to be compared, the image to be compared is obtained relative to the multiple images. The maximum similarity among the multiple feature similarities of the target image; the average or maximum value of the multiple maximum similarities corresponding to the multiple images to be compared is determined as the identity similarity.

可选地，上述目标图像包括多张；多张目标图像包含同一目标人员；上述身份相似度确定模块，还可以配置成：针对每张目标图像，获取该目标图像相对于多张待比对图像的多个特征相似度的相似度平均值；将多张目标图像对应的多个相似度平均值的最大值，确定为身份相似度。Optionally, the aforementioned target image includes multiple images; the multiple target images contain the same target person; the aforementioned identity similarity determination module can also be configured to: for each target image, obtain the target image relative to the multiple images to be compared The similarity average value of the multiple feature similarities of. The maximum value of the multiple similarity average values corresponding to multiple target images is determined as the identity similarity.

可选地，上述目标图像包括多张；多张目标图像包含同一目标人员；上述特征比对模块，还可以配置成：针对每张目标图像，比对该目标图像与待比对图像的图像特征，得到该目标图像对应的特征相似度；上述身份相似度确定模块，还可以配置成：将多张目标图像对应的特征相似度中的相似度最大值，确定为目标人员和已知身份类型人员的身份相似度。Optionally, the target image includes multiple images; the multiple target images contain the same target person; the feature comparison module may also be configured to: for each target image, compare the image features of the target image with the image to be compared , To obtain the feature similarity corresponding to the target image; the identity similarity determination module can also be configured to: determine the maximum similarity among the feature similarities corresponding to multiple target images as the target person and the person with a known identity type The similarity of identities.

可选地，上述包含有目标人员的目标图像预先存储在目标人员的第一人员多维特征集中；如果目标图像包括多张，多张目标图像中，包含有同一个人员的目标图像存储在同一个第一人员多维特征集中。Optionally, the above-mentioned target image containing the target person is pre-stored in the first person multi-dimensional feature set of the target person; if the target image includes multiple images, among the multiple target images, the target image containing the same person is stored in the same The first person has a multi-dimensional feature set.

可选地，上述包含已知身份类型人员的待比对图像预先存储在已知身份类型人员的第二人员多维特征集中；如果待比对图像包括多张，多张待比对图像中，包含有同一个人员的待比对图像存储在同一个第二人员多维特征集中。Optionally, the above-mentioned image to be compared containing persons with known identity types is pre-stored in the second person multi-dimensional feature set of persons with known identity types; if the image to be compared includes multiple images, the multiple images to be compared include The images to be compared for the same person are stored in the same second person multidimensional feature set.

可选地，上述待比对图像包括多张；多张待比对图像存储在至少两个第二人员多维特征集中；至少两个第二人员多维特征集中的待比对图像所包含的已知身份类型人员的身份类型相同；目标图像包括多张；多张目标图像存储在同一个第一人员多维特征集中；上述特征比对模块，还可以配置成：针对每个第二人员多维特征集，执行下述操作：针对该第二人员多维特征集中的每个待比对图像，比较多张目标图像的图像特征与该待比对图像的图像特征，得到多张目标图像相对于该待比对图像的多个特征相似度；上述身份相似度确定模块，还可以配置成：针对每个第二人员多维特征集，执行以下操作：针对该第二 人员多维特征集中的每张待比对图像，获取该待比对图像相对于多张目标图像的多个特征相似度中的相似度最大值；将该第二人员多维特征集中多张待比对图像对应的多个相似度最大值的平均值或最大值，确定为目标人员与该第二人员多维特征集对应的已知身份类型的人员的身份相似度；将目标人员与多个第二人员多维特征集对应的已知身份类型的人员的多个身份相似度的平均值，确定为身份相似度。Optionally, the above-mentioned image to be compared includes multiple images; the multiple images to be compared are stored in at least two second person multi-dimensional feature sets; the at least two second person multi-dimensional feature sets contain known images to be compared The identity types of persons with identity types are the same; there are multiple target images; multiple target images are stored in the same multi-dimensional feature set of the first person; the above-mentioned feature comparison module can also be configured to: for each multi-dimensional feature set of the second person, Perform the following operation: For each image to be compared in the second person's multidimensional feature set, compare the image features of multiple target images with the image features of the image to be compared to obtain multiple target images relative to the image to be compared The similarity of multiple features of the image; the above identity similarity determination module can also be configured to: for each second person multi-dimensional feature set, perform the following operations: for each second person multi-dimensional feature set to be compared, Obtain the maximum similarity among multiple feature similarities of the image to be compared with respect to multiple target images; the average value of the multiple maximum similarities corresponding to the multiple images to be compared in the second person's multidimensional feature set Or the maximum value, which is determined as the identity similarity between the target person and the person of the known identity type corresponding to the second person’s multi-dimensional feature set; the target person’s identity is the person of the known identity type corresponding to the multiple second person’s multi-dimensional feature set The average of multiple identity similarities is determined as identity similarity.

可选地，上述待比对图像包括多张；多张待比对图像存储在至少两个第二人员多维特征集中；至少两个第二人员多维特征集中的待比对图像所包含的已知身份类型人员的身份类型相同；目标图像包括多张；多张目标图像存储在同一个第一人员多维特征集中；上述特征比对模块，还可以配置成：针对每个第二人员多维特征集，执行下述操作：针对每张目标图像，比较该第二人员多维特征集中的多张待比对图像的图像特征与该目标图像的图像特征，得到多张待比对图像相对于该目标图像的多个特征相似度；上述身份相似度确定模块，还可以配置成：针对每个第二人员多维特征集，执行以下操作：针对每张目标图像，获取该目标图像相对于该第二人员多维特征集中的多张待比对图像的多个特征相似度的相似度平均值；将多张目标图像对应的多个相似度平均值的最大值，确定为目标人员与该第二人员多维特征集对应的已知身份类型的人员的身份相似度；将目标人员与多个第二人员多维特征集对应的已知身份类型的人员的多个身份相似度的平均值，确定为身份相似度。Optionally, the above-mentioned image to be compared includes multiple images; the multiple images to be compared are stored in at least two second person multi-dimensional feature sets; the at least two second person multi-dimensional feature sets contain known images to be compared The identity types of persons with identity types are the same; there are multiple target images; multiple target images are stored in the same multi-dimensional feature set of the first person; the above-mentioned feature comparison module can also be configured to: for each multi-dimensional feature set of the second person, Perform the following operations: For each target image, compare the image features of the multiple images to be compared in the second person's multi-dimensional feature set with the image features of the target image to obtain the comparison of the multiple images to be compared with the target image. Multiple feature similarities; the aforementioned identity similarity determination module can also be configured to: for each second person multi-dimensional feature set, perform the following operations: For each target image, obtain the target image relative to the second person multi-dimensional feature The similarity average value of the multiple feature similarities of the multiple images to be compared in the set; the maximum value of the multiple similarity average values corresponding to the multiple target images is determined as the target person corresponding to the second person's multi-dimensional feature set The identity similarity of persons with known identity types; the average value of multiple identity similarities between the target person and persons with known identity types corresponding to the multiple second person multi-dimensional feature sets is determined as the identity similarity.

可选地，上述身份类型确定模块，还可以配置成：如果身份相似度高于预设的相似度阈值，确定目标人员的身份类型为已知身份类型。Optionally, the above-mentioned identity type determination module may also be configured to determine that the identity type of the target person is a known identity type if the identity similarity is higher than a preset similarity threshold.

可选地，上述第二人员多维特征集包括多个多维特征集组；同一多维特征集组内的第二人员多维特征集的身份类型相同；不同多维特征集组的第二人员多维特征集的身份类型不同；目标图像包括多张；多张目标图像存储在同一个第一人员多维特征集中；上述特征比对模块，还可以配置成：针对每组多维特征集组中的每个第二人员多维特征集，针对该第二人员多维特征集中的每个待比对图像，比较第一人员多维特征集中的多张目标图像的图像特征与该待比对图像的图像特征，得到第一人员多维特征集相对于该多维特征集组的多个特征相似度；上述身份相似度确定模块，还可以配置成：针对每个多维特征集组中的每个第二人员多维特征集，执行以下操作：针对该第二人员多维特征集中的每张待比对图像，获取该待比对图像相对于多张目标图像的多个特征相似度中的相似度最大值；将多张待比对图像对应的多个相似度最大值的平均值或最大值，确定为目标人员与该第二人员多维特征集对应的已知身份类型的人员的身份相似度；将目标人员与同一多维特征集组中的多个第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为目标人员与该多维特征集组的身份相似度。Optionally, the above-mentioned second person multi-dimensional feature set includes multiple multi-dimensional feature set groups; the identity types of the second person multi-dimensional feature sets in the same multi-dimensional feature set group are the same; the second person multi-dimensional feature sets of different multi-dimensional feature set groups Different identity types; target images include multiple images; multiple target images are stored in the same first person multi-dimensional feature set; the feature comparison module can also be configured to: for each second person in each multi-dimensional feature set group Multi-dimensional feature set, for each image to be compared in the multi-dimensional feature set of the second person, compare the image features of multiple target images in the multi-dimensional feature set of the first person with the image features of the image to be compared to obtain the multi-dimensional feature set of the first person The feature set has multiple feature similarities relative to the multi-dimensional feature set group; the aforementioned identity similarity determination module can also be configured to: perform the following operations for each second person's multi-dimensional feature set in each multi-dimensional feature set group: For each image to be compared in the multi-dimensional feature set of the second person, obtain the maximum similarity among the feature similarities of the image to be compared with respect to multiple target images; The average or maximum value of multiple maximum similarities is determined as the identity similarity between the target person and the person of the known identity type corresponding to the second person’s multi-dimensional feature set; The average value of the multiple identity similarities of persons with known identity types corresponding to the second person multi-dimensional feature set is determined as the identity similarity between the target person and the multi-dimensional feature set group.

可选地，上述第二人员多维特征集包括多个多维特征集组；同一多维特征集组内的第二人员多维特征集的身份类型相同；不同多维特征集组的第二人员多维特征集的身份类型不同；目标图像包括多张；多张目标图像存储在同一个第一人员多维特征集中；上述特征比对模块，还可以配置成：针对每组多维特征集组中的每个第二人员多维特征集，执行下述操作：针对每张目标图像，比较该第二人员多维特征集中的多张待比对图像的图像特征与该目标图像的图像特征，得到多张待比对图像相对于该目标图像的多个特征相似度；上述身份相似度确定模块，还可以配置成：针对每个多维特征集组中的每个第二人员多维特征集，执行以下操作：针对每张目标图像，获取该目标图像相对于该第二人员多维特征集中的多张待比对图像的多个特征相似度的相似度平均值；将多张目标图像对应的多个相似度平均值的最大值，确定为目标人员与该第二人员多维特征集对应的已知身份类型的人员的身份相似度；将目标人员与同一多维特征集组中的多个第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为目标人员与该多维特征集组的身份相似度。Optionally, the second person multi-dimensional feature set includes multiple multi-dimensional feature set groups; the identity types of the second person multi-dimensional feature sets in the same multi-dimensional feature set group are the same; the second person multi-dimensional feature sets of different multi-dimensional feature set groups Different identity types; target images include multiple images; multiple target images are stored in the same first person multi-dimensional feature set; the feature comparison module can also be configured to: for each second person in each multi-dimensional feature set group Multi-dimensional feature set, perform the following operations: For each target image, compare the image features of the multiple images to be compared in the second person's multi-dimensional feature set with the image features of the target image to obtain multiple images to be compared relative to The multiple feature similarities of the target image; the aforementioned identity similarity determination module can also be configured to: perform the following operations for each second person multi-dimensional feature set in each multi-dimensional feature set group: For each target image, Obtain the similarity average of the multiple feature similarities of the target image with respect to the multiple images to be compared in the second person's multi-dimensional feature set; determine the maximum value of the multiple similarity averages corresponding to the multiple target images Is the identity similarity between the target person and the person of the known identity type corresponding to the second person’s multi-dimensional feature set; the target person’s identity type is the known identity type corresponding to multiple second person multi-dimensional feature sets in the same multi-dimensional feature set group The average value of the plurality of said identity similarities of persons is determined as the identity similarity of the target person and the multi-dimensional feature set group.

可选地，上述身份类型确定模块，还可以配置成：从多组多维特征集组对应的身份相似度中，选择最高的身份相似度，将最高的身份相似度对应的多维特征集组的身份类型，确定为目标人员的身份类型；或者，针对每组多维特征集组，判断该多维特征集组对应的身份相似度，是否高于该多维特征集组对应的相似度阈值；如果该多维特征集组对应的身份相似度高于相似度阈值，将该多维特征集组对应的身份类型，确定为目标人员的身份类型。Optionally, the aforementioned identity type determination module can also be configured to select the highest identity similarity from the identity similarities corresponding to the multiple sets of multi-dimensional feature set groups, and assign the highest identity similarity to the identity of the multi-dimensional feature set group Type, determined as the identity type of the target person; or, for each multi-dimensional feature set group, determine whether the identity similarity corresponding to the multi-dimensional feature set group is higher than the similarity threshold corresponding to the multi-dimensional feature set group; if the multi-dimensional feature set group The identity similarity corresponding to the set group is higher than the similarity threshold, and the identity type corresponding to the multi-dimensional feature set group is determined as the identity type of the target person.

可选地，上述第一人员多维特征集中的目标图像设置有采集时间；第一人员多维特征集中包括多张目标图像；上述装置还包括工作时间确定模块，配置成：根据每张目标图像对应的特征相似度，以及目标图像的采集时间，确定目标人员的工作时间。Optionally, the target image in the first person multi-dimensional feature set is set with a collection time; the first person multi-dimensional feature set includes multiple target images; the foregoing device further includes a working time determining module configured to: The feature similarity and the acquisition time of the target image determine the working time of the target person.

可选地，上述工作时间确定模块，还可以配置成：针对每张目标图像，判断该目标图像的特征相似度是否高于预设的相似度阈值；如果该目标图像的特征相似度高于预设的相似度阈值，确定该目标图像对应的采集时间属于目标人员的工作时间；将属于目标人员的工作时间的采集时间所组成的时间段，确定为目标人员的工作时间。Optionally, the above-mentioned working time determination module may also be configured to: for each target image, determine whether the feature similarity of the target image is higher than a preset similarity threshold; if the feature similarity of the target image is higher than the preset similarity threshold; Set the similarity threshold to determine that the acquisition time corresponding to the target image belongs to the working time of the target person; the time period composed of the acquisition time belonging to the working time of the target person is determined as the working time of the target person.

上述装置还包括人员数量确定模块，配置成：获取预设区域范围内的目标人员的工作时间；根据目标人员的工作时间，确定预设区域范围内，在指定时间点处于工作状态的目标人员的数量。The above device also includes a personnel quantity determination module, configured to: obtain the working hours of the target personnel in the preset area; according to the working hours of the target personnel, determine the work time of the target personnel in the preset area within the specified time point. quantity.

可选地，上述目标人员包括多个；多个目标人员的第一人员多维特征集属于同一预设区域范围；上述装置还包括筛选模块，配置成：根据每个目标人员的第一人员多维特征集所对应的身份类型，从预设区域范围内的目标人员中筛选得到指定人员。Optionally, the aforementioned target person includes multiple; the first person multi-dimensional feature set of the multiple target persons belongs to the same preset area range; the aforementioned device further includes a screening module configured to: according to the first person multi-dimensional feature of each target person The identity type corresponding to the set is selected from the target personnel within the preset area to obtain the designated personnel.

本实施例还提供一种电子系统，电子系统包括：处理设备和存储装置；存储装置上存储有计算机程序，计算机程序在被处理设备运行时执行如上述确定人员身份类型的方法。This embodiment also provides an electronic system. The electronic system includes: a processing device and a storage device; the storage device stores a computer program, and the computer program executes the above-mentioned method for determining the identity type of a person when the processed device is running.

本实施例还提供一种机器可读存储介质，机器可读存储介质上存储有计算机程序，计算机程序被处理设备运行时执行如上述确定人员身份类型的步骤。This embodiment also provides a machine-readable storage medium, and a computer program is stored on the machine-readable storage medium. When the computer program is run by a processing device, the steps of determining the identity type of a person as described above are executed.

所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的系统和装置的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。Those skilled in the art can clearly understand that, for the convenience and conciseness of the description, the specific working process of the system and device described above can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here.

另外，在本申请实施例的描述中，除非另有明确的规定和限定，术语“安装”、“相连”、“连接”应做广义理解，例如，可以是固定连接，也可以是可拆卸连接，或一体地连接；可以是机械连接，也可以是电连接；可以是直接相连，也可以通过中间媒介间接相连，可以是两个元件内部的连通。对于本领域技术人员而言，可以具体情况理解上述术语在本申请中的具体含义。In addition, in the description of the embodiments of the present application, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it may be a fixed connection or a detachable connection. , Or integrally connected; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the internal communication between two components. For those skilled in the art, the specific meaning of the above-mentioned terms in this application can be understood under specific circumstances.

所述功能如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本申请各个实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solution of the present application essentially or the part that contributes to the existing technology or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks or optical disks and other media that can store program codes. .

最后应说明的是：以上实施例，仅为本申请的具体实施方式，用以说明本申请的技术方案，而非对其限制，本申请的保护范围并不局限于此，尽管参照前述实施例对本申请进行了详细的说明，本领域技术人员应当理解：任何熟悉本技术领域的技术人员在本申请揭露的技术范围内，其依然可以对前述实施例所记载的技术方案进行修改或可轻易想到变化，或者对其中部分技术特征进行等同替换；而这些修改、变化或者替换，并不使相应技术方案的本质脱离本申请实施例技术方案的精神和范围，都应涵盖在本申请的保护范围之内。因此，本申请的保护范围应以权利要求的保护范围为准。Finally, it should be noted that the above examples are only specific implementations of the application, to illustrate the technical solutions of the application, but not to limit it. The scope of protection of the application is not limited thereto, although referring to the foregoing examples This application has been described in detail, and those skilled in the art should understand that any person skilled in the art within the technical scope disclosed in this application can still modify the technical solutions described in the foregoing embodiments or can easily imagine Changes, or equivalent replacements of some of the technical features; and these modifications, changes or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should be covered by the scope of protection of this application Inside. Therefore, the protection scope of this application shall be subject to the protection scope of the claims.

工业实用性Industrial applicability

本申请提供了一种确定人员身份类型的方法、装置和电子系统；其中，该方法包括获取包含有目标人员的目标图像的图像特征，以及包含已知身份类型人员的待比对图像的图像特征；其中，图像特征至少包括人员的人体外形特征；待比对图像的图像特征能够体现已知身份类型人员的身份类型；比对目标图像和待比对图像的图像特征，得到特征相似度；根据特征相似度，确定目标人员和已知身份类型人员的身份相似度；根据身份相似度，确定目标人员的身份类型。该方式可以确定没有身份信息的目标人员的身份类型，且准确率较高。This application provides a method, device, and electronic system for determining the identity type of a person; wherein the method includes acquiring image features of a target image containing a target person, and image features of an image to be compared containing a person with a known identity type ; Among them, the image features include at least the human body shape features of the person; the image features of the image to be compared can reflect the identity type of the person with a known identity type; the image features of the target image and the image to be compared are compared to obtain the feature similarity; according to Feature similarity determines the identity similarity between the target person and the known identity type; according to the identity similarity, the identity type of the target person is determined. This method can determine the identity type of the target person without identity information, and the accuracy rate is high.

Claims (21)

Translated fromChinese

一种确定人员身份类型的方法，其特征在于，所述方法包括：A method for determining the identity type of a person, characterized in that the method includes:获取包含有目标人员的目标图像的图像特征，以及包含已知身份类型人员的待比对图像的图像特征；其中，所述目标图像的图像特征至少包括所述人员的人体外形特征；所述待比对图像的图像特征至少包括所述人员的人体外形特征；所述待比对图像的图像特征用于表征所述已知身份类型人员的身份类型；Obtain the image features of the target image containing the target person and the image features of the image to be compared for the person with a known identity type; wherein, the image feature of the target image includes at least the human body shape feature of the person; The image feature of the compared image includes at least the human body shape feature of the person; the image feature of the image to be compared is used to characterize the identity type of the person with the known identity type;比对所述目标图像和所述待比对图像的图像特征，得到特征相似度；Comparing image features of the target image and the image to be compared to obtain feature similarity;根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度；Determine the identity similarity between the target person and the person with the known identity type according to the feature similarity;根据所述身份相似度，确定所述目标人员的身份类型。According to the identity similarity, the identity type of the target person is determined.根据权利要求1所述的方法，其特征在于，所述待比对图像包括多张；每张所述待比对图像中包含一个所述已知身份类型人员；多张所述待比对图像中包含的所述已知身份类型人员的身份类型相同；The method according to claim 1, wherein the images to be compared include multiple images; each image to be compared includes one person with the known identity type; and multiple images to be compared The identity types of the persons with the known identity types included in the identities are the same;所述比对所述目标图像和所述待比对图像的图像特征，得到特征相似度的步骤，包括：比较每张所述目标图像的图像特征与每张所述待比对图像的图像特征，得到多个特征相似度；The step of comparing the image features of the target image and the image to be compared to obtain feature similarity includes: comparing the image feature of each target image with the image feature of each image to be compared , Get the similarity of multiple features;所述根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：根据所述多个特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度。The step of determining the identity similarity between the target person and the person with the known identity type according to the feature similarity includes: determining the target person and the known person according to the plurality of feature similarities The similarity of the identity of the personnel of the identity type.根据权利要求2所述的方法，其特征在于，所述目标图像包括多张；多张所述目标图像包含同一目标人员；The method according to claim 2, wherein the target image includes multiple images; the multiple target images include the same target person;所述根据所述多个特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：The step of determining the identity similarity between the target person and the person with the known identity type according to the multiple feature similarities includes:针对每张所述待比对图像，获取该待比对图像相对于多张所述目标图像的多个特征相似度中的相似度最大值；For each of the images to be compared, obtaining the maximum similarity among the feature similarities of the image to be compared with respect to the plurality of target images;将多张所述待比对图像对应的多个所述相似度最大值的平均值或最大值，确定为所述身份相似度。The identity similarity is determined as the average value or the maximum value of the multiple maximum similarity values corresponding to the multiple images to be compared.根据权利要求2所述的方法，其特征在于，所述目标图像包括多张；多张所述目标图像包含同一目标人员；The method according to claim 2, wherein the target image includes multiple images; the multiple target images include the same target person;所述根据所述多个特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：The step of determining the identity similarity between the target person and the person with the known identity type according to the multiple feature similarities includes:针对每张所述目标图像，获取该目标图像相对于多张所述待比对图像的多个特征相似度的相似度平均值；For each of the target images, obtaining an average similarity of the similarity of multiple features of the target image with respect to the plurality of images to be compared;将多张所述目标图像对应的多个所述相似度平均值的最大值，确定为所述身份相似度。The maximum value of the multiple average values of the similarities corresponding to the multiple target images is determined as the identity similarity.根据权利要求1所述的方法，其特征在于，所述目标图像包括多张；多张所述目标图像包含同一目标人员；The method according to claim 1, wherein the target image includes multiple images; the multiple target images include the same target person;所述比对所述目标图像和所述待比对图像的图像特征，得到特征相似度的步骤，包括：针对每张所述目标图像，比对该目标图像的图像特征与所述待比对图像的图像特征，得到该目标图像对应的特征相似度；The step of comparing the image features of the target image and the image to be compared to obtain the feature similarity includes: for each of the target images, comparing the image feature of the target image with the image to be compared The image features of the image, and the feature similarity corresponding to the target image is obtained;所述根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：将多张所述目标图像对应的特征相似度中的相似度最大值，确定为所述目标人员和所述已知身份类型人员的身份相似度。The step of determining the identity similarity between the target person and the person with the known identity type according to the feature similarity includes: obtaining the maximum similarity among the feature similarities corresponding to the multiple target images, It is determined as the identity similarity between the target person and the person with the known identity type.根据权利要求1-5任一项所述的方法，其特征在于，所述包含有目标人员的目标图像预先存储在所述目标人员的第一人员多维特征集中；The method according to any one of claims 1 to 5, wherein the target image containing the target person is pre-stored in the first person multi-dimensional feature set of the target person;如果所述目标图像包括多张，多张所述目标图像中，包含有同一个人员的目标图像存储在同一个第一人员多维特征集中。If the target image includes multiple images, among the multiple target images, the target images containing the same person are stored in the same first person multi-dimensional feature set.根据权利要求6所述的方法，其特征在于，所述包含已知身份类型人员的待比对图像预先存储在所述已知身份类型人员的第二人员多维特征集中；The method according to claim 6, wherein the to-be-compared image containing the person of the known identity type is pre-stored in the second person multidimensional feature set of the person of the known identity type;如果所述待比对图像包括多张，多张所述待比对图像中，包含有同一个人员的待比对图像存储在同一个第二人员多维特征集中。If the image to be compared includes multiple images, among the multiple images to be compared, the images to be compared containing the same person are stored in the same second person multidimensional feature set.根据权利要求7所述的方法，其特征在于，所述待比对图像包括多张；多张所述待比对图像存储在至少两个所述第二人员多维特征集中；至少两个所述第二人员多维特征集中的待比对图像所包含的所述已知身份类型人员的身份类型相同；所述目标图像包括多张；多张所述目标图像存储在同一个第一人员多维特征集中；The method according to claim 7, wherein the images to be compared include multiple images; the multiple images to be compared are stored in at least two of the second person multidimensional feature sets; at least two of the images are The images to be compared in the second person multi-dimensional feature set contain the same identity types of the persons with known identities; the target images include multiple images; the multiple target images are stored in the same first person multi-dimensional feature set ；所述比对所述目标图像和所述待比对图像的图像特征，得到特征相似度的步骤，包括：针对每个所述第二人员多维特征集，执行下述操作：针对该第二人员多维特征集中的每张所述待比对图像，比较多张所述目标图像的图像特征与该待比对图像的图像特征，得到多张所述目标图像相对于该待比对图像的多个特征相似度；The step of comparing the image features of the target image and the image to be compared to obtain the feature similarity includes: for each of the second person's multi-dimensional feature set, the following operation is performed: for the second person For each of the images to be compared in the multi-dimensional feature set, compare the image features of a plurality of the target images with the image features of the image to be compared, and obtain a plurality of the target images relative to the image to be compared. Feature similarity所述根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：The step of determining the identity similarity between the target person and the person with the known identity type according to the feature similarity includes:针对每个所述第二人员多维特征集，执行以下操作：针对该第二人员多维特征集中的每张所述待比对图像，获取该待比对图像相对于多张所述目标图像的多个特征相似度中的相似度最大值；将该第二人员多维特征集中多张所述待比对图像对应的多个所述相似度最大值的平均值或最大值，确定为所述目标人员与该第二人员多维特征集对应的已知身份类型的人员的所述身份相似度；For each of the second person's multi-dimensional feature set, perform the following operation: For each of the to-be-compared images in the second person's multi-dimensional feature set, obtain the amount of the to-be-compared image relative to the multiple of the target images. The maximum similarity in the similarity of the features; determining the average or maximum value of the maximum similarity corresponding to the multiple images to be compared in the multidimensional feature set of the second person as the target person The identity similarity of persons of known identity types corresponding to the second person multidimensional feature set;将所述目标人员与多个所述第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为所述身份相似度。The identity similarity is determined as an average value of a plurality of the identity similarities of the target person and the persons of the known identity type corresponding to the plurality of second person multidimensional feature sets.根据权利要求7所述的方法，其特征在于，所述待比对图像包括多张；多张所述待比对图像存储在至少两个所述第二人员多维特征集中；至少两个所述第二人员多维特征集中的所述待比对图像所包含的所述已知身份类型人员的身份类型相同；所述目标图像包括多张；多张所述目标图像存储在同一个第一人员多维特征集中；The method according to claim 7, wherein the images to be compared include multiple images; the multiple images to be compared are stored in at least two of the second person multidimensional feature sets; at least two of the images are The identity types of the persons with known identity types contained in the images to be compared in the second person multi-dimensional feature set are the same; the target images include multiple images; and the multiple target images are stored in the same first person multi-dimensional feature set. Feature concentration所述比对所述目标图像和所述待比对图像的图像特征，得到特征相似度的步骤，包括：针对每个所述第二人员多维特征集，执行下述操作：针对每张所述目标图像，比较该第二人员多维特征集中的多张所述待比对图像的图像特征与该目标图像的图像特征，得到多张所述待比对图像相对于该目标图像的多个特征相似度；The step of comparing the image features of the target image and the image to be compared to obtain the feature similarity includes: performing the following operation for each of the second person multidimensional feature sets: The target image, comparing the image features of the multiple images to be compared with the image features of the target image in the multidimensional feature set of the second person, to obtain multiple features of the multiple images to be compared with respect to the target image that are similar Spend;所述根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：The step of determining the identity similarity between the target person and the person with the known identity type according to the feature similarity includes:针对每个所述第二人员多维特征集，执行以下操作：针对每张所述目标图像，获取该目标图像相对于该第二人员多维特征集中的多张所述待比对图像的多个特征相似度的相似度平均值；将多张所述目标图像对应的多个所述相似度平均值的最大值，确定为所述目标人员与该第二人员多维特征集对应的已知身份类型的人员的所述身份相似度；For each of the second person's multi-dimensional feature set, perform the following operations: For each of the target images, obtain multiple features of the target image relative to the plurality of images to be compared in the second person's multi-dimensional feature set The similarity average value of the similarity; determining the maximum value of the multiple similarity average values corresponding to the multiple target images as the known identity type corresponding to the multi-dimensional feature set of the target person and the second person The similarity of the said identities of the personnel;将所述目标人员与多个所述第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为身份相似度。The average value of the multiple identity similarities between the target person and the persons with known identity types corresponding to the multiple second person multidimensional feature sets is determined as the identity similarity.根据权利要求1-9任一项所述的方法，其特征在于，根据所述身份相似度，确定所述目标人员的身份类型的步骤，包括：The method according to any one of claims 1-9, wherein the step of determining the identity type of the target person according to the identity similarity comprises:如果所述身份相似度高于预设的相似度阈值，确定所述目标人员的身份类型为所述已知身份类型。If the identity similarity is higher than a preset similarity threshold, it is determined that the identity type of the target person is the known identity type.根据权利要求7-9任一项所述的方法，其特征在于，所述第二人员多维特征集包括多个多维特征集组；同一所述多维特征集组内的第二人员多维特征集的身份类型相同；不同所述多维特征集组的第二人员多维特征集的身份类型不同；所述目标图像包括多张；多张所述目标图像存储在同一个第一人员多维特征集中；The method according to any one of claims 7-9, wherein the second person multi-dimensional feature set includes multiple multi-dimensional feature set groups; the second person multi-dimensional feature set in the same multi-dimensional feature set group The identity types are the same; the identity types of the second person multi-dimensional feature sets of different multi-dimensional feature set groups are different; the target image includes multiple images; the multiple target images are stored in the same first person multi-dimensional feature set;所述比对所述目标图像和所述待比对图像的图像特征，得到特征相似度的步骤，包括：针对每组所述多维特征集组中的每个第二人员多维特征集，针对该第二人员多维特征集中的每张所述待比对图像，比较所述第一人员多维特征集中的多张所述目标图像的图像特征与该待比对图像的图像特征，得到所述第一人员多维特征集相对于该多维特征集组的多个特征相似度；The step of comparing the image features of the target image and the image to be compared to obtain the feature similarity includes: for each second person multi-dimensional feature set in each group of the multi-dimensional feature set group, for the For each of the images to be compared in the second person's multi-dimensional feature set, compare the image features of the multiple target images in the first person's multi-dimensional feature set with the image features of the image to be compared to obtain the first The similarity of multiple features of the multi-dimensional feature set of the person with respect to the multi-dimensional feature set group;所述根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：The step of determining the identity similarity between the target person and the person with the known identity type according to the feature similarity includes:针对每个所述多维特征集组中的每个所述第二人员多维特征集，执行以下操作：For each multi-dimensional feature set of the second person in each multi-dimensional feature set group, perform the following operations:针对该第二人员多维特征集中的每张所述待比对图像，获取该待比对图像相对于多张所述目标图像的多个特征相似度中的相似度最大值；将多张所述待比对图像对应的多个所述相似度最大值的平均值或最大值，确定为所述目标人员与该第二人员多维特征集对应的已知身份类型的人员的所述身份相似度；For each of the images to be compared in the multidimensional feature set of the second person, obtain the maximum similarity among the feature similarities of the image to be compared with respect to the multiple target images; The average value or the maximum value of the multiple maximum similarity values corresponding to the image to be compared is determined as the identity similarity between the target person and the person of the known identity type corresponding to the second person multidimensional feature set;将所述目标人员与同一多维特征集组中的多个第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为目标人员与该多维特征集组的身份相似度。Determine the average value of a plurality of said identity similarities between the target person and the plurality of second person multi-dimensional feature sets in the same multi-dimensional feature set group of persons with known identity types as the target person and the multi-dimensional feature set The identity similarity of the group.根据权利要求7-9任一项所述的方法，其特征在于，所述第二人员多维特征集包括多个多维特征集组；同一所述多维特征集组内的第二人员多维特征集的身份类型相同；不同所述多维特征集组的第二人员多维特征集的身份类型不同；所述目标图像包括多张；多张所述目标图像存储在同一个第一人员多维特征集中；The method according to any one of claims 7-9, wherein the second person multi-dimensional feature set includes multiple multi-dimensional feature set groups; the second person multi-dimensional feature set in the same multi-dimensional feature set group The identity types are the same; the identity types of the second person multi-dimensional feature sets of different multi-dimensional feature set groups are different; the target image includes multiple images; the multiple target images are stored in the same first person multi-dimensional feature set;所述比对所述目标图像和所述待比对图像的图像特征，得到特征相似度的步骤，包括：针对每组多维特征集组中的每个第二人员多维特征集，执行下述操作：针对每张所述目标图像，比较该第二人员多维特征集中的多张所述待比对图像的图像特征与该目标图像的图像特征，得到多张所述待比对图像相对于该目标图像的多个特征相似度；The step of comparing the image features of the target image and the image to be compared to obtain feature similarity includes: performing the following operations for each second person multi-dimensional feature set in each multi-dimensional feature set group : For each of the target images, compare the image features of the multiple images to be compared with the image features of the target image in the second person's multi-dimensional feature set to obtain multiple images to be compared relative to the target Similarity of multiple features of the image;所述根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度的步骤，包括：The step of determining the identity similarity between the target person and the person with the known identity type according to the feature similarity includes:针对每个所述多维特征集组中的每个所述第二人员多维特征集，执行以下操作：For each multi-dimensional feature set of the second person in each multi-dimensional feature set group, perform the following operations:针对每张所述目标图像，获取该目标图像相对于该第二人员多维特征集中的多张所述待比对图像的多个特征相似度的相似度平均值；将多张所述目标图像对应的多个所述相似度平均值的最大值，确定为所述目标人员与该第二人员多维特征集对应的已知身份类型的人员的所述身份相似度；For each of the target images, obtain the average similarity of the similarity of the target image with respect to the plurality of feature similarities of the plurality of images to be compared in the second person's multi-dimensional feature set; corresponding the plurality of target images The maximum value of the average values of the plurality of similarities is determined as the identity similarity between the target person and the person of the known identity type corresponding to the second person multidimensional feature set;将所述目标人员与同一所述多维特征集组中的多个第二人员多维特征集对应的已知身份类型的人员的多个所述身份相似度的平均值，确定为目标人员与该多维特征集组的身份相似度。Determine the average of the identity similarities between the target person and the plurality of second person multi-dimensional feature sets in the same multi-dimensional feature set group as the average value of the multiple identity similarities of persons with known identity types. Identity similarity of feature set groups.根据权利要求11或12所述的方法，其特征在于，根据所述身份相似度，确定所述目标人员的身份类型的步骤，包括：The method according to claim 11 or 12, wherein the step of determining the identity type of the target person according to the identity similarity includes:从多组所述多维特征集组对应的身份相似度中，选择最高的身份相似度，将所述最高的身份相似度对应的多维特征集组的身份类型，确定为所述目标人员的身份类型；Select the highest identity similarity from the multiple sets of identity similarities corresponding to the multi-dimensional feature set groups, and determine the identity type of the multi-dimensional feature set corresponding to the highest identity similarity as the identity type of the target person ；或者，针对每组所述多维特征集组，判断该多维特征集组对应的身份相似度是否高于该多维特征集组对应的相似度阈值；如果所述多维特征集组对应的身份相似度高于所述多维特征集组对应的相似度阈值，将该多维特征集组对应的身份类型，确定为所述目标人员的身份类型。Or, for each group of the multi-dimensional feature set group, determine whether the identity similarity corresponding to the multi-dimensional feature set group is higher than the similarity threshold corresponding to the multi-dimensional feature set group; if the identity similarity corresponding to the multi-dimensional feature set group is high Based on the similarity threshold corresponding to the multi-dimensional feature set group, the identity type corresponding to the multi-dimensional feature set group is determined as the identity type of the target person.根据权利要求6所述的方法，其特征在于，所述第一人员多维特征集中的目标图像设置有采集时间；所述第一人员多维特征集中包括多张所述目标图像；The method according to claim 6, wherein the target image in the first person multi-dimensional feature set is set with a collection time; the first person multi-dimensional feature set includes a plurality of the target images;所述根据所述身份相似度，确定所述目标人员的身份类型的步骤之后，所述方法还包括：After the step of determining the identity type of the target person according to the identity similarity, the method further includes:根据每张所述目标图像对应的特征相似度，以及所述目标图像的采集时间，确定所述目标人员的工作时间。The working time of the target person is determined according to the feature similarity corresponding to each target image and the acquisition time of the target image.根据权利要求14所述的方法，其特征在于，根据每张所述目标图像对应的特征相似度，以及所述目标图像的采集时间，确定所述目标人员的工作时间的步骤，包括：The method according to claim 14, wherein the step of determining the working time of the target person according to the feature similarity corresponding to each target image and the acquisition time of the target image comprises:针对每张所述目标图像，判断该目标图像的特征相似度是否高于预设的相似度阈值；如果所述目标图像的特征相似度高于所述预设的相似度阈值，确定该目标图像对应的采集时间属于所述目标人员的工作时间；For each target image, determine whether the feature similarity of the target image is higher than the preset similarity threshold; if the feature similarity of the target image is higher than the preset similarity threshold, determine the target image The corresponding collection time belongs to the working time of the target person;将属于所述目标人员的工作时间的采集时间所组成的时间段，确定为所述目标人员的工作时间。The time period composed of the collection time belonging to the working time of the target person is determined as the working time of the target person.根据权利要求14所述的方法，其特征在于，根据每张所述目标图像对应的特征相似度，以及所述目标图像的采集时间，确定所述目标人员的工作时间的步骤之后，所述方法还包括：The method according to claim 14, wherein after the step of determining the working time of the target person according to the feature similarity corresponding to each target image and the acquisition time of the target image, the method Also includes:获取预设区域范围内的目标人员的工作时间；Obtain the working hours of the target personnel within the preset area;根据所述目标人员的工作时间，确定所述预设区域范围内，在指定时间点处于工作状态的目标人员的数量。According to the working hours of the target personnel, the number of target personnel who are in the working state at a specified time point within the preset area is determined.根据权利要求6所述的方法，其特征在于，所述目标人员包括多个；多个所述目标人员的第一人员多维特征集属于同一预设区域范围；The method according to claim 6, wherein the target person includes a plurality of; the first person multi-dimensional feature sets of the plurality of target persons belong to the same preset area range;根据所述身份相似度，确定所述目标人员的身份类型的步骤之后，所述方法还包括：After the step of determining the identity type of the target person according to the identity similarity, the method further includes:根据每个所述目标人员的第一人员多维特征集所对应的身份类型，从所述预设区域范围内的目标人员中筛选得到指定人员。According to the identity type corresponding to the first person multi-dimensional feature set of each target person, a designated person is selected from the target persons within the preset area.根据权利要求1所述的方法，其特征在于，所述目标图像为未知身份类型的人员的多维特征集中的图像；所述方法还包括：The method according to claim 1, wherein the target image is an image in a multi-dimensional feature set of a person with an unknown identity type; the method further comprises:获取所述未知身份类型的人员的图像特征；Acquiring the image characteristics of the person with the unknown identity type;获取已知身份类型的人员图像；Obtain images of persons with known identity types;提取所述已知身份类型的人员的图像的图像特征，将所述已知身份类型的人员的图像特征与所述未知身份类型的人员的图像的图像特征进行比对，获得所述未知身份类型的人员与所述已知身份类型的人员的相似度；Extract the image features of the image of the person with the known identity type, and compare the image features of the person with the known identity type with the image features of the image of the person with the unknown identity type to obtain the unknown identity type The similarity between the personnel of and the personnel of the known identity type;如果所述未知身份类型的人员与所述已知身份类型的人员的相似度大于预设阈值，确定所述未知身份类型的人员与所述已知身份类型的人员的身份类型相同。If the similarity between the person of the unknown identity type and the person of the known identity type is greater than a preset threshold, it is determined that the identity type of the person of the unknown identity type is the same as that of the person of the known identity type.一种确定人员身份类型的装置，其特征在于，所述装置包括：A device for determining the identity type of a person, characterized in that the device comprises:特征获取模块，配置成获取包含有目标人员的目标图像的图像特征，以及包含已知身份类型人员的待比对图像的图像特征；其中，所述图像特征至少包括所述人员的人体外形特征；所述待比对图像的图像特征能够体现所述已知身份类型人员的身份类型；The feature acquisition module is configured to acquire the image features of the target image containing the target person and the image features of the image to be compared that contains the person with a known identity type; wherein the image feature includes at least the human body shape feature of the person; The image feature of the image to be compared can reflect the identity type of the known identity type;特征比对模块，配置成比对所述目标图像和所述待比对图像的图像特征，得到特征相似度；The feature comparison module is configured to compare the image features of the target image and the image to be compared to obtain feature similarity;身份相似度确定模块，配置成根据所述特征相似度，确定所述目标人员和所述已知身份类型人员的身份相似度；An identity similarity determination module, configured to determine the identity similarity between the target person and the person with the known identity type according to the feature similarity;身份类型确定模块，配置成根据所述身份相似度，确定所述目标人员的身份类型。The identity type determining module is configured to determine the identity type of the target person according to the identity similarity.一种电子系统，其特征在于，所述电子系统包括：处理设备和存储装置；An electronic system, characterized in that, the electronic system includes: a processing device and a storage device;所述存储装置上存储有计算机程序，所述计算机程序在被所述处理设备运行时执行如权利要求1-18任一项所述的确定人员身份类型的方法。A computer program is stored on the storage device, and the computer program executes the method for determining a person's identity type according to any one of claims 1-18 when the computer program is run by the processing device.一种机器可读存储介质，所述机器可读存储介质上存储有计算机程序，其特征在于，所述计算机程序被处理设备运行时执行如权利要求1-18任一项所述的确定人员身份类型的步骤。A machine-readable storage medium having a computer program stored on the machine-readable storage medium, wherein the computer program is executed when the computer program is run by a processing device to determine the identity of a person according to any one of claims 1-18 Type of steps.

Images