技术领域technical field
本发明涉及多媒体技术领域，具体而言，本发明涉及一种图像查重的方法及装置。The present invention relates to the field of multimedia technology, in particular, the present invention relates to a method and device for image plagiarism checking.
背景技术Background technique
随着信息技术的发展，多媒体技术也随之发展，各种类型的网站应运而生，一些用户或者网站管理者将经常上传一些图像至该网站，以供其它用户下载以及查看。With the development of information technology, multimedia technology also develops thereupon, and various types of websites emerge as the times require. Some users or website managers will often upload some images to the website for other users to download and view.
因此，网站会接收到大量的上传图像，但是这些上传图像中有很多图像为重复图像或者为相似度很高的图像，当网站根据用户观看量对图像进行排名，以推荐给用户时，由于这些图像中存在大量重复图像或者相似度很高的图像，将导致网站对图像排名的准确度较低，并且推荐给用户的图像的准确度也较低，并且由于这些图像中存在大量重复图像或者相似度很高的图像，也不利于用户查找观看这些图像，从而导致用户的体验度较低。Therefore, the website will receive a large number of uploaded images, but many of these uploaded images are repeated images or images with high similarity. There are a large number of repeated images or images with high similarity in the images, which will lead to low accuracy of the website's image ranking, and the accuracy of the images recommended to users is also low, and because there are a large number of repeated images or similar images in these images High-quality images are not conducive to users to find and watch these images, resulting in low user experience.
发明内容Contents of the invention
为克服上述技术问题或者至少部分地解决上述技术问题，特提出以下技术方案：In order to overcome the above-mentioned technical problems or at least partially solve the above-mentioned technical problems, the following technical solutions are proposed:
本发明的实施例根据一个方面，提供了一种图像查重的方法，包括：Embodiments of the present invention provide a method for image plagiarism checking according to one aspect, including:
获取待查重的图像，并将该待查重的图像输入预设的特征提取模型，得到该待查重的图像对应的深度特征；Obtaining an image to be checked for duplicates, and inputting the image to be checked for duplicates into a preset feature extraction model to obtain a depth feature corresponding to the image to be checked for duplicates;
其中，预设的特征提取模型是通过对深度卷积神经网络训练得到的。Among them, the preset feature extraction model is obtained by training a deep convolutional neural network.
将该待查重的图像对应的深度特征进行图像特征池化处理，得到池化处理后的该待查重的图像对应的深度特征；The depth feature corresponding to the image to be checked for repetition is subjected to image feature pooling processing, and the depth feature corresponding to the image to be checked for repetition after the pooling processing is obtained;
根据池化处理后的该待查重的图像对应的深度特征，进行图像查重。Image plagiarism check is performed according to the depth feature corresponding to the image to be checked after pooling processing.
进一步地，将该待查重的图像输入预设的特征提取模型，得到该待查重的图像对应的深度特征的步骤之前，还包括：Further, before the step of inputting the image to be checked into the preset feature extraction model, and obtaining the depth feature corresponding to the image to be checked, it also includes:
将该待查重的图像进行图像预处理，图像预处理包括以下至少一项：规整尺寸处理以及图片白化处理；Perform image preprocessing on the image to be checked, and the image preprocessing includes at least one of the following: regular size processing and image whitening processing;
其中，将该待查重的图像输入预设的特征提取模型，得到该待查重的图像对应的深度特征的步骤，包括：Wherein, the step of inputting the image to be checked into a preset feature extraction model to obtain the depth feature corresponding to the image to be checked includes:
将图像预处理后的该待查重的图像输入预设的特征提取模型，得到该待查重图像对应的深度特征。The pre-processed image to be checked is input into a preset feature extraction model to obtain the depth feature corresponding to the image to be checked.
具体地，根据池化处理后的该待查重的图像对应的深度特征，进行图像查重的步骤，包括：Specifically, according to the depth feature corresponding to the image to be checked after the pooling process, the step of image checking includes:
根据待查重的图像对应的深度特征，并通过乘积量化Product Quantization，确定待查重的图像对应的深度特征索引；According to the depth feature corresponding to the image to be checked, and quantify Product Quantization by product, determine the depth feature index corresponding to the image to be checked;
根据待查重的图像对应的深度特征索引，进行图像查重。According to the depth feature index corresponding to the image to be checked, the image is checked for duplicate.
具体地，图像查重的方式，包括：Specifically, the methods of image plagiarism check include:
判断各个图像分别对应的深度特征索引是否存在相同；Determine whether the depth feature indexes corresponding to each image are the same;
若存在相同的深度特征索引，则确定相同的深度特征索引对应的各个图像重复。If the same depth feature index exists, it is determined that each image corresponding to the same depth feature index is repeated.
进一步地，从重复的各个图像中，确定待删除的图像，并删除该待删除的图像。Further, an image to be deleted is determined from the repeated images, and the image to be deleted is deleted.
本发明的实施例根据另一个方面，还提供了一种图像查重的装置，包括：According to another aspect, the embodiments of the present invention also provide an image plagiarism checking device, including:
获取模块，用于获取待查重的图像；An acquisition module, configured to acquire images to be checked;
输入模块，用于将获取模块获取到的待查重的图像输入预设的特征提取模型，得到该待查重的图像对应的深度特征；The input module is used to input the image to be checked for repetition obtained by the acquisition module into a preset feature extraction model to obtain the depth feature corresponding to the image to be checked for repetition;
其中，预设的特征提取模型是通过对深度卷积神经网络训练得到的。Among them, the preset feature extraction model is obtained by training a deep convolutional neural network.
图像特征池化模块，用于将该待查重的图像对应的深度特征进行图像特征池化处理，得到池化处理后的该待查重的图像对应的深度特征；The image feature pooling module is used to perform image feature pooling processing on the depth feature corresponding to the image to be checked for repetition, and obtain the depth feature corresponding to the image to be checked for repetition after the pooling process;
图像查重模块，用于根据图像特征池化模块池化处理后的该待查重的图像对应的深度特征，进行图像查重。The image plagiarism checking module is configured to perform image plagiarism checking according to the depth features corresponding to the image to be checked after being pooled by the image feature pooling module.
进一步地，装置还包括：图像预处理模块；Further, the device also includes: an image preprocessing module;
图像预处理模块，用于将该待查重的图像进行图像预处理，图像预处理包括以下至少一项：规整尺寸处理以及图片白化处理；An image preprocessing module, which is used to perform image preprocessing on the image to be checked for duplication, and the image preprocessing includes at least one of the following: regular size processing and image whitening processing;
输入模块，具体用于将图像预处理模块图像预处理后的该待查重的图像输入预设的特征提取模型，得到该待查重图像对应的深度特征。The input module is specifically used to input the image pre-processed by the image preprocessing module into the preset feature extraction model to obtain the depth feature corresponding to the image to be checked.
具体地，图像查重模块包括：确定单元、图像查重单元；Specifically, the image duplicate check module includes: a determination unit, an image duplicate check unit;
确定单元，用于根据待查重的图像对应的深度特征，并通过乘积量化ProductQuantization，确定待查重的图像对应的深度特征索引；The determination unit is used to determine the depth feature index corresponding to the image to be checked according to the depth feature corresponding to the image to be checked, and quantify ProductQuantization by product;
图像查重单元，用于根据确定单元确定的待查重的图像对应的深度特征索引，进行图像查重。The image plagiarism checking unit is configured to perform image plagiarism checking according to the depth feature index corresponding to the image to be checked for plagiarism determined by the determining unit.
具体地，图像查重模块，具体用于判断各个图像分别对应的深度特征索引是否存在相同；Specifically, the image duplicate checking module is specifically used to judge whether the corresponding depth feature indexes of each image are the same;
图像查重模块，具体还用于当存在相同的深度特征索引时，确定相同的深度特征索引对应的各个图像重复。The image duplication checking module is also specifically used to determine the repetition of each image corresponding to the same depth feature index when the same depth feature index exists.
进一步地，装置还包括：确定模块、删除模块；Further, the device also includes: a determination module and a deletion module;
确定模块，用于从重复的各个图像中，确定待删除的图像；A determining module, configured to determine the image to be deleted from among the repeated images;
删除模块，用于删除该待删除的图像。The deletion module is used to delete the image to be deleted.
本发明的实施例根据又一个方面，还提供了一种计算机可读存储介质，其特征在于，计算机可读存储介质上存储有计算机程序，该程序被处理器执行时实现上述图像查重的方法。According to another aspect, the embodiment of the present invention also provides a computer-readable storage medium, which is characterized in that a computer program is stored on the computer-readable storage medium, and when the program is executed by a processor, the above-mentioned image plagiarism checking method is realized .
本发明的实施例根据又一个方面，还提供了一种计算设备，包括：处理器、存储器、通信接口和通信总线，处理器、存储器和通信接口通过通信总线完成相互间的通信；According to yet another aspect, the embodiments of the present invention also provide a computing device, including: a processor, a memory, a communication interface, and a communication bus, and the processor, the memory, and the communication interface complete mutual communication through the communication bus;
存储器用于存放至少一可执行指令，可执行指令使处理器执行上述的图像查重的方法对应的操作。The memory is used to store at least one executable instruction, and the executable instruction causes the processor to execute the operation corresponding to the above-mentioned image plagiarism checking method.
本发明提供了一种图像查重的方法及装置，本发明获取待查重的图像，并将该待查重的图像输入预设的特征提取模型，得到该待查重的图像对应的深度特征，然后将该待查重的图像对应的深度特征进行图像特征池化处理，得到池化处理后的该待查重的图像对应的深度特征，然后根据所述池化处理后的该待查重的图像对应的深度特征，进行图像查重。即本发明通过对图像进行查重，例如对已上传的图像进行查重，能够确定已上传的图像中的重复图像或者相似度很高的图像，从而可以提高网站对图像排名的准确度，并且由于对已上传的图像进行查重，因此降低了重复图像以及相似度较高的图像的概率，当用户查找图像时，能够更加准确地查找到所需图像，进而可以提升用户的体验度。The present invention provides a method and device for image plagiarism checking. The present invention acquires an image to be checked for plagiarism, and inputs the image to be checked for plagiarism into a preset feature extraction model to obtain the depth feature corresponding to the image to be checked for plagiarism. , and then perform image feature pooling processing on the depth feature corresponding to the image to be checked to obtain the depth feature corresponding to the image to be checked after the pooling process, and then according to the depth feature corresponding to the image to be checked after the pooling process The depth features corresponding to the image are used to check the image for plagiarism. That is to say, the present invention can determine duplicate images or images with high similarity in the uploaded images by performing duplicate checking on images, such as uploaded images, thereby improving the accuracy of website ranking of images, and Since uploaded images are checked for plagiarism, the probability of repeated images and images with high similarity is reduced. When users search for images, they can find the desired images more accurately, thereby improving user experience.
本发明附加的方面和优点将在下面的描述中部分给出，这些将从下面的描述中变得明显，或通过本发明的实践了解到。Additional aspects and advantages of the invention will be set forth in part in the description which follows, and will become apparent from the description, or may be learned by practice of the invention.
附图说明Description of drawings
本发明上述的和/或附加的方面和优点从下面结合附图对实施例的描述中将变得明显和容易理解，其中：The above and/or additional aspects and advantages of the present invention will become apparent and easy to understand from the following description of the embodiments in conjunction with the accompanying drawings, wherein:
图1为本发明实施例的一种图像查重的方法流程图；FIG. 1 is a flow chart of a method for image plagiarism checking according to an embodiment of the present invention;
图2为本发明实施例的三种进行池化的方式示意图；FIG. 2 is a schematic diagram of three pooling methods according to an embodiment of the present invention;
图3为本发明实施例中的一种图像查重的装置结构示意图；Fig. 3 is a schematic structural diagram of an image plagiarism checking device in an embodiment of the present invention;
图4为本发明实施例中的另一种图像查重的装置结构示意图。Fig. 4 is a schematic structural diagram of another image plagiarism checking device in an embodiment of the present invention.
具体实施方式Detailed ways
下面详细描述本发明的实施例，所述实施例的示例在附图中示出，其中自始至终相同或类似的标号表示相同或类似的元件或具有相同或类似功能的元件。下面通过参考附图描述的实施例是示例性的，仅用于解释本发明，而不能解释为对本发明的限制。Embodiments of the present invention are described in detail below, examples of which are shown in the drawings, wherein the same or similar reference numerals designate the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the figures are exemplary only for explaining the present invention and should not be construed as limiting the present invention.
本技术领域技术人员可以理解，除非特意声明，这里使用的单数形式“一”、“一个”、“所述”和“该”也可包括复数形式。应该进一步理解的是，本发明的说明书中使用的措辞“包括”是指存在所述特征、整数、步骤、操作、元件和/或组件，但是并不排除存在或添加一个或多个其他特征、整数、步骤、操作、元件、组件和/或它们的组。应该理解，当我们称元件被“连接”或“耦接”到另一元件时，它可以直接连接或耦接到其他元件，或者也可以存在中间元件。此外，这里使用的“连接”或“耦接”可以包括无线连接或无线耦接。这里使用的措辞“和/或”包括一个或更多个相关联的列出项的全部或任一单元和全部组合。Those skilled in the art will understand that unless otherwise stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the word "comprising" used in the description of the present invention refers to the presence of said features, integers, steps, operations, elements and/or components, but does not exclude the presence or addition of one or more other features, Integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Additionally, "connected" or "coupled" as used herein may include wireless connection or wireless coupling. The expression "and/or" used herein includes all or any elements and all combinations of one or more associated listed items.
本技术领域技术人员可以理解，除非另外定义，这里使用的所有术语(包括技术术语和科学术语)，具有与本发明所属领域中的普通技术人员的一般理解相同的意义。还应该理解的是，诸如通用字典中定义的那些术语，应该被理解为具有与现有技术的上下文中的意义一致的意义，并且除非像这里一样被特定定义，否则不会用理想化或过于正式的含义来解释。Those skilled in the art can understand that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this invention belongs. It should also be understood that terms, such as those defined in commonly used dictionaries, should be understood to have meanings consistent with their meaning in the context of the prior art, and unless specifically defined as herein, are not intended to be idealized or overly Formal meaning to explain.
本技术领域技术人员可以理解，这里所使用的“终端”、“终端设备”既包括无线信号接收器的设备，其仅具备无发射能力的无线信号接收器的设备，又包括接收和发射硬件的设备，其具有能够在双向通信链路上，进行双向通信的接收和发射硬件的设备。这种设备可以包括：蜂窝或其他通信设备，其具有单线路显示器或多线路显示器或没有多线路显示器的蜂窝或其他通信设备；PCS(Personal Communications Service，个人通信系统)，其可以组合语音、数据处理、传真和/或数据通信能力；PDA(Personal Digital Assistant，个人数字助理)，其可以包括射频接收器、寻呼机、互联网/内联网访问、网络浏览器、记事本、日历和/或GPS(Global Positioning System，全球定位系统)接收器；常规膝上型和/或掌上型计算机或其他设备，其具有和/或包括射频接收器的常规膝上型和/或掌上型计算机或其他设备。这里所使用的“终端”、“终端设备”可以是便携式、可运输、安装在交通工具(航空、海运和/或陆地)中的，或者适合于和/或配置为在本地运行，和/或以分布形式，运行在地球和/或空间的任何其他位置运行。这里所使用的“终端”、“终端设备”还可以是通信终端、上网终端、音乐/视频播放终端，例如可以是PDA、MID(Mobile Internet Device，移动互联网设备)和/或具有音乐/视频播放功能的移动电话，也可以是智能电视、机顶盒等设备。Those skilled in the art can understand that the "terminal" and "terminal equipment" used here not only include wireless signal receiver equipment, which only has wireless signal receiver equipment without transmission capabilities, but also include receiving and transmitting hardware. A device having receive and transmit hardware capable of bi-directional communication over a bi-directional communication link. Such equipment may include: cellular or other communication equipment, which has a single-line display or a multi-line display or a cellular or other communication equipment without a multi-line display; PCS (Personal Communications Service, personal communication system), which can combine voice, data Processing, facsimile and/or data communication capabilities; PDA (Personal Digital Assistant, Personal Digital Assistant), which may include radio frequency receiver, pager, Internet/Intranet access, web browser, notepad, calendar and/or GPS (Global Positioning System (Global Positioning System) receiver; a conventional laptop and/or palmtop computer or other device having and/or including a radio frequency receiver. As used herein, a "terminal", "terminal device" may be portable, transportable, installed in a vehicle (air, sea, and/or land), or adapted and/or configured to operate locally, and/or In distributed form, the operation operates at any other location on Earth and/or in space. The "terminal" and "terminal equipment" used here can also be communication terminals, Internet terminals, music/video playback terminals, such as PDAs, MIDs (Mobile Internet Devices, mobile Internet devices) and/or with music/video playback terminals. Functional mobile phones, smart TVs, set-top boxes and other devices.
实施例一Embodiment one
本发明实施例提供了一种图像查重的方法，如图1所示，包括：The embodiment of the present invention provides a method for image plagiarism checking, as shown in Figure 1, comprising:
步骤101、获取待查重的图像。Step 101, acquiring images to be checked for plagiarism.
步骤102、将该待查重的图像输入预设的特征提取模型，得到该待查重的图像对应的深度特征。Step 102: Input the image to be checked for plagiarism into a preset feature extraction model to obtain the depth feature corresponding to the image to be checked for plagiarism.
其中，预设的特征提取模型是通过对深度卷积神经网络训练得到的。Among them, the preset feature extraction model is obtained by training a deep convolutional neural network.
例如，通过2千万的素材图像，共21000类别对该深度卷积神经进行训练，得到该预设的特征提取模型。For example, through 20 million material images, a total of 21,000 categories are used to train the deep convolutional neural network to obtain the preset feature extraction model.
对于本发明实施例，将各个关键帧输入训练后的深度卷积神经网络，得到该各个关键帧中每个关键帧所属21000类别中每类中的概率；或者输出该关键帧对应的预设维数的表征，该表征可以用于表征该帧图像对应的应用场景，例如，室内、室外、太阳以及天空等。For the embodiment of the present invention, each key frame is input into the deep convolutional neural network after training, and the probability in each of the 21000 categories to which each key frame belongs in each key frame is obtained; or the preset dimension corresponding to the key frame is output The representation of the data can be used to represent the application scene corresponding to the frame image, for example, indoor, outdoor, sun, sky, etc.
步骤103、将该待查重的图像对应的深度特征进行图像特征池化处理，得到池化处理后的该待查重的图像对应的深度特征。Step 103 : Perform image feature pooling processing on the depth features corresponding to the image to be checked for plagiarism, and obtain the depth feature corresponding to the image to be checked for plagiarism after pooling.
对于本发明实施例，池化是在卷积特征提取的基础上，对每个卷积特征进行取平均等，继续缩小隐藏节点对于的卷积特征维数。For the embodiment of the present invention, pooling is to average each convolution feature on the basis of convolution feature extraction, and continue to reduce the convolution feature dimension of the hidden node pair.
对于本发明实施例，一个图像区域有用的特征极有可能在另一个区域同样适用。因此，为了描述大的图像，一个很自然的想法就是对不同位置的特征进行聚合统计，例如，人们可以计算图像一个区域上的某个特定特征的平均值(或最大值)。这些概要统计特征不仅具有低得多的维度(相比使用所有提取得到的特征)，同时还会改善结果(不容易过拟合)。这种聚合的操作就叫做池化(pooling)。For the embodiments of the present invention, features that are useful in one image region are likely to be applicable in another region as well. Therefore, to describe large images, a natural idea is to aggregate statistics on features at different locations, for example, one can calculate the average value (or maximum value) of a certain feature over a region of the image. These summary statistical features not only have much lower dimensionality (compared to using all extracted features), but also improve the results (less overfitting). This aggregation operation is called pooling.
对于本发明实施例，池化可以包括：1)mean-pooling，即对邻域内特征点只求平均，对背景保留更好；max-pooling，即对邻域内特征点取最大，对纹理提取更好；3)Stochastic-pooling，介于两者之间，通过对像素点按照数值大小赋予概率，再按照概率进行亚采样。For the embodiment of the present invention, pooling may include: 1) mean-pooling, that is, only average the feature points in the neighborhood, and better preserve the background; max-pooling, that is, take the maximum of the feature points in the neighborhood, and extract more Good; 3) Stochastic-pooling, between the two, by assigning a probability to the pixel according to the value, and then subsampling according to the probability.
其中，特征提取的误差主要来自两个方面：(1)邻域大小受限造成的估计值方差增大；(2)卷积层参数误差造成估计均值的偏移。一般来说，mean-pooling能减小第一种误差，更多的保留图像的背景信息，max-pooling能减小第二种误差，更多的保留纹理信息。在平均意义上，与mean-pooling近似，在局部意义上，则服从max-pooling的准则。其中上述三种池化的方式如图2所示。Among them, the error of feature extraction mainly comes from two aspects: (1) the increase in the variance of the estimated value caused by the limited size of the neighborhood; (2) the deviation of the estimated mean value caused by the error of the convolutional layer parameters. Generally speaking, mean-pooling can reduce the first error and retain more background information of the image, while max-pooling can reduce the second error and retain more texture information. In the average sense, it is similar to mean-pooling, and in the local sense, it obeys the max-pooling criterion. The above three pooling methods are shown in Figure 2.
步骤104、根据池化处理后的该待查重的图像对应的深度特征，进行图像查重。Step 104 , perform image plagiarism check according to the depth feature corresponding to the image to be checked for plagiarism after pooling processing.
对于本发明实施例，通过待查重图像的特征信息，确定已上线的图像中是否存在与该待查重图像的特征信息关联度较高的图像，以实现图像查重。For the embodiment of the present invention, by using the feature information of the image to be checked for plagiarism, it is determined whether there is an image with a high degree of correlation with the feature information of the image to be checked for plagiarism among the online images, so as to realize image plagiarism checking.
本发明实施例提供了一种图像查重的方法，本发明实施例获取待查重的图像，并将该待查重的图像输入预设的特征提取模型，得到该待查重的图像对应的深度特征，然后将该待查重的图像对应的深度特征进行图像特征池化处理，得到池化处理后的该待查重的图像对应的深度特征，然后根据所述池化处理后的该待查重的图像对应的深度特征，进行图像查重。即本发明实施例通过对图像进行查重，例如对已上传的图像进行查重，能够确定已上传的图像中的重复图像或者相似度很高的图像，从而可以提高网站对图像排名的准确度，并且由于对已上传的图像进行查重，因此降低了重复图像以及相似度较高的图像的概率，当用户查找图像时，能够更加准确地查找到所需图像，进而可以提升用户的体验度。An embodiment of the present invention provides a method for image plagiarism checking. The embodiment of the present invention acquires an image to be checked for plagiarism, and inputs the image to be checked for plagiarism into a preset feature extraction model to obtain the image corresponding to the image to be checked for plagiarism. Depth features, and then perform image feature pooling processing on the depth features corresponding to the image to be checked, and obtain the depth features corresponding to the image to be checked after the pooling process, and then according to the pooling process. Check the depth features corresponding to the image for plagiarism, and perform image plagiarism check. That is to say, the embodiment of the present invention can determine duplicate images or images with high similarity in the uploaded images by performing duplicate checking on images, such as uploaded images, thereby improving the accuracy of website ranking of images , and because the uploaded images are checked for duplicates, the probability of repeated images and images with high similarity is reduced. When users search for images, they can find the desired images more accurately, which in turn can improve user experience .
实施例二Embodiment two
本发明实施例的另一种可能的实现方式，在实施例一所示的基础上，还包括实施例二所示的操作，其中，Another possible implementation of the embodiment of the present invention, on the basis of the first embodiment, also includes the operation shown in the second embodiment, wherein,
步骤102之前还包括：之前，还包括：将该待查重的图像进行图像预处理。Before step 102, it also includes: before, it also includes: performing image preprocessing on the image to be checked.
其中，图像预处理包括以下至少一项：规整尺寸处理以及图片白化处理。Wherein, the image preprocessing includes at least one of the following: regular size processing and image whitening processing.
对于本发明实施例，通过将各个关键帧分别进行规整尺寸处理以及图片白化处理，以提高各个关键帧图像的鲁棒性。For the embodiment of the present invention, the robustness of each key frame image is improved by subjecting each key frame to regular size processing and picture whitening processing.
对于本发明实施例，对图像进行规整尺寸处理为通过采样的方式对图像进行规整尺寸处理，例如，将图像中扣取五个大块，包括从该图像的中间以及四角分别进行扣取。For the embodiment of the present invention, performing regular size processing on the image is to perform regular size processing on the image by sampling, for example, extracting five large blocks from the image, including extracting from the middle and four corners of the image respectively.
对于本发明实施例，图像最终成像会受环境照明强度、物体反射、拍摄相机等多因素的影响。为了能够图像的中包含的那些不受外界影响的恒定信息，对图像进行图像白化处理。一般为了去除这些因素的影响，我们将它的像素值转化成零均值和单位方差。因此首先通过公式一以及公式二计算原始灰度图像P的像素平均值μ和方差值δ2。For the embodiment of the present invention, the final imaging of the image will be affected by many factors such as ambient lighting intensity, object reflection, and shooting camera. In order to obtain the constant information contained in the image that is not affected by the outside world, the image is whitened. Generally, in order to remove the influence of these factors, we convert its pixel values into zero mean and unit variance. Therefore, firstly, calculate the pixel average value μ and variance value δ2 of the original grayscale image P through Formula 1 and Formula 2.
其中公式一为：The first formula is:
公式二为：Formula two is:
然后，将使用μ和δ来对原始灰度图像的每个像素值进行转化：对于彩色图像，在三个通道分别计算μ和δ2，然后根据公式三分别进行像素转化。Then, μ and δ will be used to convert each pixel value of the original grayscale image: for color images, μ and δ2 are calculated in the three channels, and then the pixel conversion is performed according to formula three.
其中公式三：Among them, formula three:
步骤102包括：将图像预处理后的该待查重的图像输入预设的特征提取模型，得到该待查重图像对应的深度特征。Step 102 includes: inputting the pre-processed image to a preset feature extraction model to obtain the depth feature corresponding to the image to be checked.
实施例三Embodiment three
本发明实施例的另一种可能的实现方式，在实施例三所示的基础上，还包括实施例四所示的操作，其中，Another possible implementation of the embodiment of the present invention, on the basis of the third embodiment, also includes the operation shown in the fourth embodiment, wherein,
步骤104包括：根据待查重的图像对应的深度特征，并通过乘积量化ProductQuantization，确定待查重的图像对应的深度特征索引；根据待查重的图像对应的深度特征索引，进行图像查重。Step 104 includes: determining the depth feature index corresponding to the image to be checked for plagiarism according to the depth feature corresponding to the image to be checked for plagiarism by product quantization ProductQuantization; performing image plagiarism checking according to the depth feature index corresponding to the image to be checked for plagiarism.
对于本发明实施例，乘积量化Product Quantization包括两个过程特征的分组量化过程和类别的笛卡尔积过程。假设有一个数据集，那么K-means为将给定类别数目K，目标函数是所有样本到类中心的距离和最小值，迭代计算优化目标函数，得到K个类中心和每个样本所属的类别。目标函数不变，乘积量化的做法为：For the embodiment of the present invention, Product Quantization includes a grouping quantization process of two process features and a Cartesian product process of categories. Assuming that there is a data set, then K-means will give the number of categories K, the objective function is the distance and the minimum value of all samples to the class center, iteratively calculate and optimize the objective function, and get K class centers and the category to which each sample belongs . The objective function remains unchanged, and the method of product quantization is as follows:
(1)数据集为K个类别，每个样本以一个vector的形式表示，维数为d，将vector的各个分量分成m组。(1) The data set consists of K categories, each sample is expressed in the form of a vector, the dimension is d, and each component of the vector is divided into m groups.
(2)将所有vector的某组分量作为数据集，采用k-means算法得到个类中心，运行m次k-means算法，则每组都有个类中心，记这个类中心为一个集合。(2) Use a certain component of all vectors as a data set, and use the k-means algorithm to obtain class center, run the k-means algorithm m times, then each group has A class center, remember this A class center is a set.
(3)将上述得到的m个集合做笛卡尔积，就得到整个数据集的类中心了。(3) Do the Cartesian product of the m sets obtained above to obtain the class center of the entire data set.
对于本发明实施例，将处理后的待查重图像的特征信息，通过乘积量化ProductQuantization，得到待查重图像的图像特征索引，其中该待查重图像的图像特征索引为待查重图像与特征索引之间的对应关系。For the embodiment of the present invention, the feature information of the image to be checked for duplicates after processing is quantified by ProductQuantization to obtain the image feature index of the image to be checked for duplicates, wherein the image feature index of the image to be checked for duplicates is the image to be checked for duplicates and the feature Correspondence between indexes.
例如，待查重图像包括图像1、图像2以及图像3，分别对应的索引值为001、002、003，图像1、图像2以及图像3分别对应的图像特征的索引值为1、2、1。For example, the image to be checked includes image 1, image 2, and image 3, and the corresponding index values are 001, 002, and 003 respectively, and the index values of image features corresponding to image 1, image 2, and image 3 are 1, 2, and 1, respectively. .
其中，图像查重的方式，包括：判断各个图像分别对应的深度特征索引是否存在相同；若存在相同的深度特征索引，则确定相同的深度特征索引对应的各个图像重复。Wherein, the image duplication checking method includes: judging whether the depth feature indexes corresponding to the respective images are the same; if there is the same depth feature index, determining that the images corresponding to the same depth feature index are repeated.
对于本发明实施例，若两个图像分别对应的图像特征索引相同，则表征这两个图像为重复图像。For the embodiment of the present invention, if the image feature indexes corresponding to the two images are the same, the two images are characterized as repeated images.
例如，待查重图像包括图像1、图像2以及图像3，分别对应的索引值为001、002、003，图像1、图像2以及图像3分别对应的图像特征的索引值为1、2、1，由于图像1以及图像2对应的图像特征的索引值均为1(两个不同的图像对应的图像特征的索引值相同)因此图像1以及图像2为重复图像。For example, the image to be checked includes image 1, image 2, and image 3, and the corresponding index values are 001, 002, and 003 respectively, and the index values of image features corresponding to image 1, image 2, and image 3 are 1, 2, and 1, respectively. , because the index values of image features corresponding to image 1 and image 2 are both 1 (the index values of image features corresponding to two different images are the same), so image 1 and image 2 are repeated images.
进一步地，从重复的各个图像中，确定待删除的图像，并删除该待删除的图像。Further, an image to be deleted is determined from the repeated images, and the image to be deleted is deleted.
对于本发明实施例，若已上线的图像中存在多个重复图像，则从该多个重复图像中选择待删除的图像，并删除。For the embodiment of the present invention, if there are multiple repeated images among the online images, the image to be deleted is selected from the multiple repeated images and deleted.
对于本发明实施例，按照预设原则，从重复的各个图像中，确定待删除的图像，其中预设原则包括以下至少一项：图像的清晰度、图像的发布时间、图像的观看量以及图像的下载量。For the embodiment of the present invention, the images to be deleted are determined from the repeated images according to preset principles, wherein the preset principles include at least one of the following: the clarity of the image, the release time of the image, the amount of viewing of the image, and the downloads.
例如，已上线的图像中包括两个重复的图像，包括：图像1以及图像3，其中图像1的下载量为100，图像2的下载量为1200，则待删除的图像为图像1。For example, the online image includes two duplicate images, including image 1 and image 3, where image 1 has 100 downloads and image 2 has 1200 downloads, then the image to be deleted is image 1.
对于本发明实施例，通过从重复的各个图像中，确定待删除的图像，并删除该待删除的图像，当用户从已上线的图像中下载对应的图像时，能够准确的确定并下载待下载的图像，从而可以降低已上线图像中图像的重复率，进而可以提高查找待下载图像的准确度，提升用户的体验度。For the embodiment of the present invention, by determining the image to be deleted from the repeated images and deleting the image to be deleted, when the user downloads the corresponding image from the online images, the user can accurately determine and download the image to be downloaded. images, which can reduce the repetition rate of images in the online images, thereby improving the accuracy of finding images to be downloaded and improving user experience.
对于本发明实施例，通过从重复的各个图像中，确定待删除的图像，并删除该待删除的图像，即将重复的图像从数据库中删除，能够降低数据库中的存储量，节省存储空间。For the embodiment of the present invention, by determining the image to be deleted from the repeated images and deleting the image to be deleted, that is, deleting the repeated image from the database, the storage capacity in the database can be reduced and the storage space can be saved.
本发明实施例提供了一种图像处理的装置，如图3所示，该装置包括：获取模块31、输入模块32、图像特征池化模块33、图像查重模块34；其中，An embodiment of the present invention provides an image processing device. As shown in FIG. 3 , the device includes: an acquisition module 31, an input module 32, an image feature pooling module 33, and an image duplication checking module 34; wherein,
获取模块31，用于获取待查重的图像。An acquisition module 31, configured to acquire images to be checked for plagiarism.
输入模块32，用于将获取模块31获取到的待查重的图像输入预设的特征提取模型，得到该待查重的图像对应的深度特征。The input module 32 is configured to input the image to be checked acquired by the acquisition module 31 into a preset feature extraction model to obtain a depth feature corresponding to the image to be checked.
其中，预设的特征提取模型是通过对深度卷积神经网络训练得到的。Among them, the preset feature extraction model is obtained by training a deep convolutional neural network.
图像特征池化模块33，用于将该待查重的图像对应的深度特征进行图像特征池化处理，得到池化处理后的该待查重的图像对应的深度特征。The image feature pooling module 33 is configured to perform image feature pooling processing on the depth features corresponding to the image to be checked to obtain the depth feature corresponding to the image to be checked after pooling.
图像查重模块34，用于根据图像特征池化模块33池化处理后的该待查重的图像对应的深度特征，进行图像查重。The image plagiarism checking module 34 is used to perform image plagiarism checking according to the depth features corresponding to the image to be checked after being pooled by the image feature pooling module 33.
进一步地，如图4所示，该装置还包括：图像预处理模块41。Further, as shown in FIG. 4 , the device further includes: an image preprocessing module 41 .
图像预处理模块41，用于将该待查重的图像进行图像预处理。The image preprocessing module 41 is configured to perform image preprocessing on the image to be checked for plagiarism.
其中，图像预处理包括以下至少一项：规整尺寸处理以及图片白化处理，Wherein, the image preprocessing includes at least one of the following: regular size processing and image whitening processing,
输入模块32，具体用于将图像预处理模块41图像预处理后的该待查重的图像输入预设的特征提取模型，得到该待查重图像对应的深度特征。The input module 32 is specifically configured to input the image pre-processed by the image preprocessing module 41 into the preset feature extraction model to obtain the depth feature corresponding to the image to be checked.
进一步地，如图4所示，图像查重模块34包括：确定单元341、图像查重单元342。Further, as shown in FIG. 4 , the image plagiarism checking module 34 includes: a determining unit 341 and an image plagiarism checking unit 342 .
确定单元341，用于根据待查重的图像对应的深度特征，并通过乘积量化ProductQuantization，确定待查重的图像对应的深度特征索引。The determination unit 341 is configured to determine the depth feature index corresponding to the image to be checked based on the depth feature corresponding to the image to be checked and quantify ProductQuantization by product.
图像查重单元342，用于根据确定单元341确定的待查重的图像对应的深度特征索引，进行图像查重。The image plagiarism checking unit 342 is configured to perform image plagiarism checking according to the depth feature index corresponding to the image to be checked for plagiarism determined by the determining unit 341 .
图像查重模块34，具体用于判断各个图像分别对应的深度特征索引是否存在相同。The image plagiarism checking module 34 is specifically used for judging whether the corresponding depth feature indexes of each image are the same.
图像查重模块34，具体还用于当存在相同的深度特征索引时，确定相同的深度特征索引对应的各个图像重复。The image plagiarism checking module 34 is also specifically used to determine the repetition of each image corresponding to the same depth feature index when the same depth feature index exists.
进一步地，如图4所示，该装置还包括：确定模块42、删除模块43。Further, as shown in FIG. 4 , the device further includes: a determination module 42 and a deletion module 43 .
确定模块42，用于从重复的各个图像中，确定待删除的图像。The determination module 42 is configured to determine the image to be deleted from the repeated images.
删除模块43，用于删除该待删除的图像。The deletion module 43 is configured to delete the image to be deleted.
本发明实施例提供了一种图像查重的装置，本发明实施例获取待查重的图像，并将该待查重的图像输入预设的特征提取模型，得到该待查重的图像对应的深度特征，然后将该待查重的图像对应的深度特征进行图像特征池化处理，得到池化处理后的该待查重的图像对应的深度特征，然后根据所述池化处理后的该待查重的图像对应的深度特征，进行图像查重。即本发明实施例通过对图像进行查重，例如对已上传的图像进行查重，能够确定已上传的图像中的重复图像或者相似度很高的图像，从而可以提高网站对图像排名的准确度，并且由于对已上传的图像进行查重，因此降低了重复图像以及相似度较高的图像的概率，当用户查找图像时，能够更加准确地查找到所需图像，进而可以提升用户的体验度。An embodiment of the present invention provides a device for image plagiarism checking. The embodiment of the present invention acquires an image to be checked for plagiarism, and inputs the image to be checked for plagiarism into a preset feature extraction model to obtain the image corresponding to the image to be checked for plagiarism. Depth features, and then perform image feature pooling processing on the depth features corresponding to the image to be checked, and obtain the depth features corresponding to the image to be checked after the pooling process, and then according to the pooling process. Check the depth features corresponding to the image for plagiarism, and perform image plagiarism check. That is to say, the embodiment of the present invention can determine duplicate images or images with high similarity in the uploaded images by performing duplicate checking on images, such as uploaded images, thereby improving the accuracy of website ranking of images , and because the uploaded images are checked for duplicates, the probability of repeated images and images with high similarity is reduced. When users search for images, they can find the desired images more accurately, which in turn can improve user experience .
本发明实施例提供的图像查重的装置可以实现上述提供的方法实施例，具体功能实现请参见方法实施例中的说明，在此不再赘述。The image plagiarism checking device provided by the embodiment of the present invention can implement the method embodiment provided above. For specific function implementation, please refer to the description in the method embodiment, and details will not be repeated here.
本发明实施例提供了一种计算机可读存储介质，计算机可读存储介质上存储有计算机程序，该程序被处理器执行时实现上述图像查重的方法。An embodiment of the present invention provides a computer-readable storage medium. A computer program is stored on the computer-readable storage medium. When the program is executed by a processor, the above-mentioned image plagiarism checking method is realized.
本发明实施例提供了一种计算机可读存储介质，本发明实施例获取待查重的图像，并将该待查重的图像输入预设的特征提取模型，得到该待查重的图像对应的深度特征，然后将该待查重的图像对应的深度特征进行图像特征池化处理，得到池化处理后的该待查重的图像对应的深度特征，然后根据所述池化处理后的该待查重的图像对应的深度特征，进行图像查重。即本发明实施例通过对图像进行查重，例如对已上传的图像进行查重，能够确定已上传的图像中的重复图像或者相似度很高的图像，从而可以提高网站对图像排名的准确度，并且由于对已上传的图像进行查重，因此降低了重复图像以及相似度较高的图像的概率，当用户查找图像时，能够更加准确地查找到所需图像，进而可以提升用户的体验度。An embodiment of the present invention provides a computer-readable storage medium. The embodiment of the present invention acquires an image to be checked for plagiarism, and inputs the image to be checked for plagiarism into a preset feature extraction model to obtain an image corresponding to the image to be checked for plagiarism. Depth features, and then perform image feature pooling processing on the depth features corresponding to the image to be checked, and obtain the depth features corresponding to the image to be checked after the pooling process, and then according to the pooling process. Check the depth features corresponding to the image for plagiarism, and perform image plagiarism check. That is to say, the embodiment of the present invention can determine duplicate images or images with high similarity in the uploaded images by performing duplicate checking on images, such as uploaded images, thereby improving the accuracy of website ranking of images , and because the uploaded images are checked for duplicates, the probability of repeated images and images with high similarity is reduced. When users search for images, they can find the desired images more accurately, which in turn can improve user experience .
本发明实施例提供的计算机可读存储介质可以实现上述提供的方法实施例，具体功能实现请参见方法实施例中的说明，在此不再赘述。The computer-readable storage medium provided by the embodiments of the present invention can implement the method embodiments provided above. For specific function realization, please refer to the description in the method embodiments, and details are not repeated here.
本发明实施例提供了一种计算设备，包括：处理器、存储器、通信接口和通信总线，处理器、存储器和通信接口通过通信总线完成相互间的通信；An embodiment of the present invention provides a computing device, including: a processor, a memory, a communication interface, and a communication bus, and the processor, the memory, and the communication interface complete mutual communication through the communication bus;
存储器用于存放至少一可执行指令，可执行指令使处理器执行如上述的图像查重的方法对应的操作。The memory is used to store at least one executable instruction, and the executable instruction causes the processor to perform operations corresponding to the above-mentioned image plagiarism checking method.
本发明实施例提供了一种计算机设备，本发明实施例获取待查重的图像，并将该待查重的图像输入预设的特征提取模型，得到该待查重的图像对应的深度特征，然后将该待查重的图像对应的深度特征进行图像特征池化处理，得到池化处理后的该待查重的图像对应的深度特征，然后根据所述池化处理后的该待查重的图像对应的深度特征，进行图像查重。即本发明实施例通过对图像进行查重，例如对已上传的图像进行查重，能够确定已上传的图像中的重复图像或者相似度很高的图像，从而可以提高网站对图像排名的准确度，并且由于对已上传的图像进行查重，因此降低了重复图像以及相似度较高的图像的概率，当用户查找图像时，能够更加准确地查找到所需图像，进而可以提升用户的体验度。An embodiment of the present invention provides a computer device. The embodiment of the present invention acquires an image to be checked for plagiarism, and inputs the image to be checked for plausibility into a preset feature extraction model to obtain a depth feature corresponding to the image to be checked for plausibility. Then carry out the image feature pooling process to the depth feature corresponding to the image to be checked for repetition, obtain the depth feature corresponding to the image to be checked for repetition after the pooling process, and then according to the depth feature corresponding to the image to be checked for repetition after the pooling process The depth feature corresponding to the image is used for image plagiarism check. That is to say, the embodiment of the present invention can determine duplicate images or images with high similarity in the uploaded images by performing duplicate checking on images, such as uploaded images, thereby improving the accuracy of website ranking of images , and because the uploaded images are checked for duplicates, the probability of repeated images and images with high similarity is reduced. When users search for images, they can find the desired images more accurately, which in turn can improve user experience .
本发明实施例提供的计算机设备可以实现上述提供的方法实施例，具体功能实现请参见方法实施例中的说明，在此不再赘述。The computer equipment provided by the embodiments of the present invention can implement the method embodiments provided above. For specific function realization, please refer to the description in the method embodiments, and details are not repeated here.
本技术领域技术人员可以理解，本发明包括涉及用于执行本申请中所述操作中的一项或多项的设备。这些设备可以为所需的目的而专门设计和制造，或者也可以包括通用计算机中的已知设备。这些设备具有存储在其内的计算机程序，这些计算机程序选择性地激活或重构。这样的计算机程序可以被存储在设备(例如，计算机)可读介质中或者存储在适于存储电子指令并分别耦联到总线的任何类型的介质中，所述计算机可读介质包括但不限于任何类型的盘(包括软盘、硬盘、光盘、CD-ROM、和磁光盘)、ROM(Read-Only Memory，只读存储器)、RAM(Random Access Memory，随即存储器)、EPROM(Erasable ProgrammableRead-Only Memory，可擦写可编程只读存储器)、EEPROM(Electrically ErasableProgrammable Read-Only Memory，电可擦可编程只读存储器)、闪存、磁性卡片或光线卡片。也就是，可读介质包括由设备(例如，计算机)以能够读的形式存储或传输信息的任何介质。Those skilled in the art will appreciate that the present invention includes devices related to performing one or more of the operations described in this application. These devices may be specially designed and fabricated for the required purposes, or they may include known devices found in general purpose computers. These devices have computer programs stored therein that are selectively activated or reconfigured. Such a computer program can be stored in a device (e.g., computer) readable medium, including but not limited to any type of medium suitable for storing electronic instructions and respectively coupled to a bus. Types of disks (including floppy disks, hard disks, CDs, CD-ROMs, and magneto-optical disks), ROM (Read-Only Memory, read-only memory), RAM (Random Access Memory, random access memory), EPROM (Erasable Programmable Read-Only Memory, Erasable Programmable Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory, Electrically Erasable Programmable Read-Only Memory), flash memory, magnetic card or optical card. That is, a readable medium includes any medium that stores or transmits information in a form readable by a device (eg, a computer).
本技术领域技术人员可以理解，可以用计算机程序指令来实现这些结构图和/或框图和/或流图中的每个框以及这些结构图和/或框图和/或流图中的框的组合。本技术领域技术人员可以理解，可以将这些计算机程序指令提供给通用计算机、专业计算机或其他可编程数据处理方法的处理器来实现，从而通过计算机或其他可编程数据处理方法的处理器来执行本发明公开的结构图和/或框图和/或流图的框或多个框中指定的方案。Those skilled in the art will understand that computer program instructions can be used to implement each block in these structural diagrams and/or block diagrams and/or flow diagrams and combinations of blocks in these structural diagrams and/or block diagrams and/or flow diagrams . Those skilled in the art can understand that these computer program instructions can be provided to general-purpose computers, professional computers, or processors of other programmable data processing methods for implementation, so that the computer or processors of other programmable data processing methods can execute the present invention. A scheme specified in a block or blocks of a structure diagram and/or a block diagram and/or a flow diagram of the invention disclosure.
本技术领域技术人员可以理解，本发明中已经讨论过的各种操作、方法、流程中的步骤、措施、方案可以被交替、更改、组合或删除。进一步地，具有本发明中已经讨论过的各种操作、方法、流程中的其他步骤、措施、方案也可以被交替、更改、重排、分解、组合或删除。进一步地，现有技术中的具有与本发明中公开的各种操作、方法、流程中的步骤、措施、方案也可以被交替、更改、重排、分解、组合或删除。Those skilled in the art can understand that the various operations, methods, and steps, measures, and solutions in the processes discussed in the present invention can be replaced, changed, combined, or deleted. Further, other steps, measures, and schemes in the various operations, methods, and processes that have been discussed in the present invention may also be replaced, changed, rearranged, decomposed, combined, or deleted. Further, steps, measures, and schemes in the prior art that have operations, methods, and processes disclosed in the present invention can also be alternated, changed, rearranged, decomposed, combined, or deleted.
以上所述仅是本发明的部分实施方式，应当指出，对于本技术领域的普通技术人员来说，在不脱离本发明原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也应视为本发明的保护范围。The above descriptions are only part of the embodiments of the present invention. It should be pointed out that those skilled in the art can make some improvements and modifications without departing from the principles of the present invention. It should be regarded as the protection scope of the present invention.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711009829.8ACN107844541A (en) | 2017-10-25 | 2017-10-25 | Image duplicate checking method and device |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711009829.8ACN107844541A (en) | 2017-10-25 | 2017-10-25 | Image duplicate checking method and device |
| Publication Number | Publication Date |
|---|---|
| CN107844541Atrue CN107844541A (en) | 2018-03-27 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711009829.8APendingCN107844541A (en) | 2017-10-25 | 2017-10-25 | Image duplicate checking method and device |
| Country | Link |
|---|---|
| CN (1) | CN107844541A (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109902194A (en)* | 2019-01-18 | 2019-06-18 | 平安科技（深圳）有限公司 | Image duplication checking method and related equipment based on neural network |
| CN110362697A (en)* | 2019-07-05 | 2019-10-22 | 武汉莱博信息技术有限公司 | Image duplicate checking method, equipment and storage medium based on convolutional neural networks model |
| CN110929069A (en)* | 2019-10-14 | 2020-03-27 | 广西壮族自治区科学技术情报研究所 | Science and technology project duplicate checking method for big data matching calculation based on image partition |
| CN115661472A (en)* | 2022-11-15 | 2023-01-31 | 中国平安财产保险股份有限公司 | Image duplicate checking method and device, computer equipment and storage medium |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103744931A (en)* | 2013-12-30 | 2014-04-23 | 中国科学院深圳先进技术研究院 | Method and system for searching image |
| CN104484418A (en)* | 2014-12-17 | 2015-04-01 | 中国科学技术大学 | Characteristic quantification method and system based on double resolution factors |
| US20160247290A1 (en)* | 2015-02-23 | 2016-08-25 | Mitsubishi Electric Research Laboratories, Inc. | Method for Labeling Images of Street Scenes |
| CN106503729A (en)* | 2016-09-29 | 2017-03-15 | 天津大学 | A kind of generation method of the image convolution feature based on top layer weights |
| CN107209864A (en)* | 2015-01-27 | 2017-09-26 | 北京市商汤科技开发有限公司 | Face identification method and system |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103744931A (en)* | 2013-12-30 | 2014-04-23 | 中国科学院深圳先进技术研究院 | Method and system for searching image |
| CN104484418A (en)* | 2014-12-17 | 2015-04-01 | 中国科学技术大学 | Characteristic quantification method and system based on double resolution factors |
| CN107209864A (en)* | 2015-01-27 | 2017-09-26 | 北京市商汤科技开发有限公司 | Face identification method and system |
| US20160247290A1 (en)* | 2015-02-23 | 2016-08-25 | Mitsubishi Electric Research Laboratories, Inc. | Method for Labeling Images of Street Scenes |
| CN106503729A (en)* | 2016-09-29 | 2017-03-15 | 天津大学 | A kind of generation method of the image convolution feature based on top layer weights |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109902194A (en)* | 2019-01-18 | 2019-06-18 | 平安科技（深圳）有限公司 | Image duplication checking method and related equipment based on neural network |
| CN110362697A (en)* | 2019-07-05 | 2019-10-22 | 武汉莱博信息技术有限公司 | Image duplicate checking method, equipment and storage medium based on convolutional neural networks model |
| CN110929069A (en)* | 2019-10-14 | 2020-03-27 | 广西壮族自治区科学技术情报研究所 | Science and technology project duplicate checking method for big data matching calculation based on image partition |
| CN115661472A (en)* | 2022-11-15 | 2023-01-31 | 中国平安财产保险股份有限公司 | Image duplicate checking method and device, computer equipment and storage medium |
| Publication | Publication Date | Title |
|---|---|---|
| CN116580257B (en) | Feature fusion model training and sample retrieval method, device and computer equipment | |
| CN107665261B (en) | Video duplicate checking method and device | |
| CN108062780B (en) | Image compression method and device | |
| CN112418292B (en) | Image quality evaluation method, device, computer equipment and storage medium | |
| CN113658122B (en) | Image quality evaluation method, device, storage medium and electronic device | |
| WO2020221278A1 (en) | Video classification method and model training method and apparatus thereof, and electronic device | |
| CN109344893B (en) | A kind of image classification method based on mobile terminal | |
| US9697592B1 (en) | Computational-complexity adaptive method and system for transferring low dynamic range image to high dynamic range image | |
| CN110826567B (en) | Optical character recognition method, device, equipment and storage medium | |
| CN107844541A (en) | Image duplicate checking method and device | |
| CN112149699B (en) | Method and device for generating model and method and device for identifying image | |
| CN107705805B (en) | Audio duplicate checking method and device | |
| CN113743277A (en) | A kind of short video classification method and system, equipment and storage medium | |
| CN110097010A (en) | Picture and text detection method, device, server and storage medium | |
| WO2022178975A1 (en) | Noise field-based image noise reduction method and apparatus, device, and storage medium | |
| US20230419452A1 (en) | Method and device for correcting image on basis of compression quality of image in electronic device | |
| CN115630236A (en) | Global fast retrieval positioning method of passive remote sensing image, storage medium and equipment | |
| CN112668595B (en) | Image processing method, device, equipment and storage medium | |
| CN113742525A (en) | Self-supervision video hash learning method, system, electronic equipment and storage medium | |
| CN112307243A (en) | Method and apparatus for retrieving images | |
| US9875386B2 (en) | System and method for randomized point set geometry verification for image identification | |
| CN111935487B (en) | Image compression method and system based on video stream detection | |
| CN115953849A (en) | Training method of living body detection model, living body detection method and system | |
| CN110765304A (en) | Image processing method, image processing device, electronic equipment and computer readable medium | |
| CN113742524A (en) | Video quick retrieval method and system and video quick recommendation method |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | Application publication date:20180327 | |
| RJ01 | Rejection of invention patent application after publication |