CN111814511A - Identification method based on customized data processing - Google Patents

Abstract

The invention relates to a recognition method based on customized data processing, which comprises the steps of providing a recognition device based on customized data processing, introducing a neural network recognition mechanism to carry out on-site recognition on infant face types including a tired face type, a hungry face type, a painful face type and a water-deficient face type, and wirelessly transmitting the recognition result to a parent mobile terminal.

Description

Translated fromChinese

基于定制数据处理的识别方法Identification method based on customized data processing

技术领域technical field

本发明涉及医疗器械领域，尤其涉及一种基于定制数据处理的识别方法。The invention relates to the field of medical devices, in particular to an identification method based on customized data processing.

背景技术Background technique

智能家具，是在现代时尚家具的基础上，将组合智能、电子智能、机械智能、物联智能巧妙的融入家具产品当中，使家具智能化、国际化、时尚化，使家居生活更加便捷、舒适，是新贵生活方式重要组成部分，是未来国际家具的发展潮流和趋势。Smart furniture, on the basis of modern and fashionable furniture, ingeniously integrates combined intelligence, electronic intelligence, mechanical intelligence, and IoT intelligence into furniture products, making furniture intelligent, international, and fashionable, and making home life more convenient and comfortable. , is an important part of the upstart lifestyle, and is the future development trend and trend of international furniture.

智能家具产品组合智能打破了传统家具的组合模式，充分发挥了用户的主观创造性(DIY)，外形尺寸、组合模式不再是家具生产厂家说的算，而是用户根据个人的喜好和家庭空间的实际情况，自由组合自由搭配。人们通过将家具功能拆分、单元化加工，每一个单元就是一件产品，产品可以通过产品的排列叠加进行组合。Intelligent furniture product combination intelligently breaks the traditional furniture combination mode, and gives full play to the user's subjective creativity (DIY). The size and combination mode are no longer determined by the furniture manufacturer, but by the user according to personal preferences and family space. The actual situation, free combination and free collocation. People divide and unitize furniture functions, each unit is a product, and products can be combined through the arrangement and superposition of products.

发明内容SUMMARY OF THE INVENTION

本发明需要具备以下几处关键的发明点：The present invention needs to have the following key invention points:

(1)在针对性的图像处理的基础上，引入神经网络识别机制对包括疲倦脸部类型、饥饿脸部类型、痛苦脸部类型和缺水脸部类型的婴儿面部类型进行现场识别，并将识别结果无线发送给家长的移动终端，从而方便家长及时采取后续应对措施；(1) On the basis of targeted image processing, a neural network recognition mechanism was introduced to identify infant face types including tired face types, hungry face types, painful faces and dehydrated faces, and identified The identification result is wirelessly sent to the parent's mobile terminal, so that it is convenient for the parent to take follow-up measures in a timely manner;

(2)在图像复原处理的基础上，识别图像中各个冗余像素点和各个非冗余像素点，基于图像中的各个非冗余像素点的数量与单个像素点占用的位数的乘积计算图像的数据量；(2) On the basis of image restoration processing, identify each redundant pixel point and each non-redundant pixel point in the image, and calculate based on the product of the number of each non-redundant pixel point in the image and the number of bits occupied by a single pixel point The amount of data of the image;

(3)为了避免整个图像处理机制的运算量过大，在检测到待处理图像的数据量超限时，将运算量复杂的无缩放变换模糊设备从图像处理部件中剔除。(3) In order to avoid the excessive calculation amount of the whole image processing mechanism, when it is detected that the data amount of the image to be processed exceeds the limit, the non-scaling transform blur device with complex calculation amount is eliminated from the image processing component.

根据本发明的一方面，提供一种基于定制数据处理的识别方法，所述方法包括提供一种基于定制数据处理的识别装置，用于引入神经网络识别机制对包括疲倦脸部类型、饥饿脸部类型、痛苦脸部类型和缺水脸部类型的婴儿面部类型进行现场识别，并将识别结果无线发送给家长的移动终端，所述基于定制数据处理的识别装置包括：According to an aspect of the present invention, there is provided a recognition method based on customized data processing, the method includes providing a recognition device based on customized data processing, for introducing a neural network recognition mechanism to identify types including tired faces, hungry faces Type, painful face type and baby face type of dehydration face type are identified on the spot, and the identification result is wirelessly sent to the parent's mobile terminal, and the identification device based on customized data processing includes:

纽扣采集设备，设置在婴儿床的上方，用于对婴儿床所在环境进行图像数据采集，以获得并输出相应的床体环境图像；The button acquisition device is arranged above the crib, and is used to collect image data of the environment where the crib is located, so as to obtain and output the corresponding bed environment image;

复原处理设备，位于婴儿床内，与所述纽扣采集设备连接，用于接收所述床体环境图像，对所述床体环境图像执行图像复原处理，以获得即时复原图像；Restoration processing equipment, located in the crib, connected to the button collection device, for receiving the bed environment image, and performing image restoration processing on the bed environment image to obtain an instant restoration image;

即时统计设备，位于婴儿床内，与所述复原处理设备连接，用于接收所述即时复原图像，对所述即时复原图像的数据量进行统计，并在统计的数据量超限时，发出第一控制信号；The instant statistics device, located in the crib, is connected with the restoration processing device, and is used for receiving the instant restoration image, making statistics on the data volume of the instant restoration image, and sending out the first count when the counted data volume exceeds the limit. control signal;

所述即时统计设备还用于在统计的数据量未超限时，发出第二控制信号；The real-time statistics device is further configured to send a second control signal when the amount of data to be counted does not exceed the limit;

在所述即时统计设备中，对所述即时复原图像的数据量进行统计包括：识别所述即时复原图像中各个冗余像素点和各个非冗余像素点，基于所述即时复原图像中的各个非冗余像素点的数量与单个像素点占用的位数的乘积计算所述即时复原图像的数据量；In the instant statistics device, performing statistics on the data amount of the instant restoration image includes: identifying each redundant pixel point and each non-redundant pixel point in the instant restoration image, and based on each redundant pixel point in the instant restoration image The product of the number of non-redundant pixel points and the number of bits occupied by a single pixel point calculates the data amount of the instant restoration image;

畸变处理设备，与所述复原处理设备连接，用于接收所述即时复原图像，对所述即时复原图像执行畸变校正处理，以获得并输出相应的畸变处理图像；Distortion processing equipment, connected to the restoration processing equipment, for receiving the instant restoration image, and performing distortion correction processing on the instant restoration image to obtain and output the corresponding distortion processing image;

曲线处理设备，与所述畸变处理设备连接，用于对接收到的畸变处理图像执行曲线处理，以获得对应的曲线处理图像；a curve processing device, connected to the distortion processing device, for performing curve processing on the received distortion processed image to obtain a corresponding curve processed image;

双边滤波设备，与所述曲线处理设备连接，用于对接收到的曲线处理图像进行双边滤波处理，以获得并输出对应的双边滤波图像；A bilateral filtering device, connected to the curve processing device, for performing bilateral filtering processing on the received curve processing image to obtain and output a corresponding bilateral filtering image;

无缩放变换模糊设备，分别与所述即时统计设备和所述双边滤波设备连接，用于接收所述双边滤波图像，并在接收到所述第一控制信号时，将所述双边滤波图像作为即时平滑图像输出。The non-scaling transformation blur device is connected to the instant statistics device and the bilateral filtering device respectively, and is used for receiving the bilateral filtering image, and when receiving the first control signal, the bilateral filtering image is used as the instant Smooth image output.

本发明的基于定制数据处理的识别装置识别准确，方便使用。由于引入神经网络识别机制对婴儿面部类型进行现场识别，并将识别结果无线发送给家长的移动终端，从而方便家长及时采取后续应对措施。The identification device based on customized data processing of the present invention has accurate identification and is convenient to use. Because the neural network recognition mechanism is introduced to recognize the baby's face type on the spot, and the recognition result is wirelessly sent to the parent's mobile terminal, so that the parent can take follow-up measures in a timely manner.

附图说明Description of drawings

以下将结合附图对本发明的实施方案进行描述，其中：Embodiments of the present invention will be described below with reference to the accompanying drawings, wherein:

图1为根据本发明实施方案示出的基于定制数据处理的识别装置所应用的婴儿床的部件拆解图。FIG. 1 is a disassembled view of components of a crib to which the identification device based on customized data processing is applied according to an embodiment of the present invention.

具体实施方式Detailed ways

下面将参照附图对本发明的实施方案进行详细说明。Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

婴儿床是指给婴幼儿使用的床。现在的婴儿床款式多种多样，功能和价格也是相差也很大，选择时要注意安全和实用相结合的原则。婴儿床在使用中也存在一些安全隐患，父母应给予重视，及时排除隐患。婴儿床油漆是否符合安全标准，关键要看是否含有重金属和甲醛等有害成分。A crib is a bed for an infant. There are various styles of cribs, and the functions and prices are also very different. When choosing, pay attention to the principle of combining safety and practicality. There are also some hidden dangers in the use of cribs, parents should pay attention to them and eliminate the hidden dangers in time. Whether the crib paint meets safety standards depends on whether it contains harmful ingredients such as heavy metals and formaldehyde.

实木婴儿床目前国内有两种松木做原料，新西兰松和樟子松，新西兰松木材较软易刨光，而樟子松较硬，光洁度要差一些。Solid wood cribs currently have two kinds of pine as raw materials in China, New Zealand pine and Pinus sylvestris. New Zealand pine is soft and easy to plan, while Pinus sylvestris is harder and the finish is worse.

新型的是以竹制的婴儿床，其优点就是：竹材比木材更坚硬密实，抗压抗弯强度更高；纹清晰、板面美观、色泽自然、竹香怡人，质感高雅气派；竹材不积尘、不结露、易清洁，避免了螨类细菌的繁殖，免去虫蛀之扰；能自动调节环境湿度并抗湿，导热系数低，具冬暖夏凉的特性；竹材具有吸收紫外线的功能，使人在室内起居时眼睛有舒适感，可预防近视等眼疾的发生和发展；吸音隔声、除低音、压残响，有效摈除杂声，还你宁静心境。其缺点就是竹的凉性较重，所以竹做的婴儿床床板最好还是用木条或木板的，其他部分用竹的较好。The new type of crib is made of bamboo. Its advantages are: bamboo is harder and denser than wood, and has higher compressive and bending strength; clear pattern, beautiful board surface, natural color, pleasant bamboo fragrance, and elegant texture; Dust accumulation, non-condensing, easy to clean, avoids the reproduction of mites and bacteria, and avoids the disturbance of insects; it can automatically adjust the environmental humidity and resist moisture, with low thermal conductivity, with the characteristics of being warm in winter and cool in summer; bamboo has the characteristics of absorbing ultraviolet rays It can make people feel comfortable when living indoors, and can prevent the occurrence and development of eye diseases such as myopia. The disadvantage is that the cooling property of bamboo is heavier, so it is better to use wooden strips or planks for the crib bed made of bamboo, and bamboo is better for other parts.

目前，婴儿床的使用仅仅限于供婴儿躺卧使用，对于旁边的家长起不到任何数据提醒作用，而随着老龄化的社会的到来，人工资源越来越显得珍贵，人工智能成为科技发展的主要方向，如何对婴儿床进行人工智能化改造，以使得能够为忙碌于其他事务的家长提供更多建设性的帮助，是目前研发的重点之一。At present, the use of cribs is only limited to the use of lying down babies, and can not provide any data reminders for the parents next to them. With the advent of an aging society, artificial resources are becoming more and more precious, and artificial intelligence has become a key factor in the development of science and technology. The main direction, how to carry out artificial intelligence transformation of the crib, so as to provide more constructive help for parents who are busy with other affairs, is one of the current research and development priorities.

为了克服上述不足，本发明搭建一种基于定制数据处理的识别方法，所述方法包括提供一种基于定制数据处理的识别装置，用于引入神经网络识别机制对包括疲倦脸部类型、饥饿脸部类型、痛苦脸部类型和缺水脸部类型的婴儿面部类型进行现场识别，并将识别结果无线发送给家长的移动终端，所述基于定制数据处理的识别装置能够有效解决相应的技术问题。In order to overcome the above deficiencies, the present invention builds a recognition method based on customized data processing, the method includes providing a recognition device based on customized data processing, for introducing a neural network recognition mechanism to identify types including tired faces, hungry faces On-site identification of infant face types, distressed face types and dehydrated face types, and wirelessly send the identification results to the parent's mobile terminal, the identification device based on customized data processing can effectively solve the corresponding technical problems.

图1为根据本发明实施方案示出的基于定制数据处理的识别装置所应用的婴儿床的部件拆解图。FIG. 1 is a disassembled view of components of a crib to which the identification device based on customized data processing is applied according to an embodiment of the present invention.

根据本发明实施方案示出的基于定制数据处理的识别装置包括：The identification device based on customized data processing shown according to an embodiment of the present invention includes:

纽扣采集设备，设置在婴儿床的上方，用于对婴儿床所在环境进行图像数据采集，以获得并输出相应的床体环境图像；The button acquisition device is arranged above the crib, and is used to collect image data of the environment where the crib is located, so as to obtain and output the corresponding bed environment image;

复原处理设备，位于婴儿床内，与所述纽扣采集设备连接，用于接收所述床体环境图像，对所述床体环境图像执行图像复原处理，以获得即时复原图像；Restoration processing equipment, located in the crib, connected to the button collection device, for receiving the bed environment image, and performing image restoration processing on the bed environment image to obtain an instant restoration image;

即时统计设备，位于婴儿床内，与所述复原处理设备连接，用于接收所述即时复原图像，对所述即时复原图像的数据量进行统计，并在统计的数据量超限时，发出第一控制信号；The instant statistics device, located in the crib, is connected with the restoration processing device, and is used for receiving the instant restoration image, making statistics on the data volume of the instant restoration image, and sending out the first count when the counted data volume exceeds the limit. control signal;

所述即时统计设备还用于在统计的数据量未超限时，发出第二控制信号；The real-time statistics device is further configured to send a second control signal when the amount of data to be counted does not exceed the limit;

在所述即时统计设备中，对所述即时复原图像的数据量进行统计包括：识别所述即时复原图像中各个冗余像素点和各个非冗余像素点，基于所述即时复原图像中的各个非冗余像素点的数量与单个像素点占用的位数的乘积计算所述即时复原图像的数据量；In the instant statistics device, performing statistics on the data amount of the instant restoration image includes: identifying each redundant pixel point and each non-redundant pixel point in the instant restoration image, and based on each redundant pixel point in the instant restoration image The product of the number of non-redundant pixel points and the number of bits occupied by a single pixel point calculates the data amount of the instant restoration image;

畸变处理设备，与所述复原处理设备连接，用于接收所述即时复原图像，对所述即时复原图像执行畸变校正处理，以获得并输出相应的畸变处理图像；Distortion processing equipment, connected to the restoration processing equipment, for receiving the instant restoration image, and performing distortion correction processing on the instant restoration image to obtain and output the corresponding distortion processing image;

曲线处理设备，与所述畸变处理设备连接，用于对接收到的畸变处理图像执行曲线处理，以获得对应的曲线处理图像；a curve processing device, connected to the distortion processing device, for performing curve processing on the received distortion processed image to obtain a corresponding curve processed image;

双边滤波设备，与所述曲线处理设备连接，用于对接收到的曲线处理图像进行双边滤波处理，以获得并输出对应的双边滤波图像；A bilateral filtering device, connected to the curve processing device, for performing bilateral filtering processing on the received curve processing image to obtain and output a corresponding bilateral filtering image;

无缩放变换模糊设备，分别与所述即时统计设备和所述双边滤波设备连接，用于接收所述双边滤波图像，并在接收到所述第一控制信号时，将所述双边滤波图像作为即时平滑图像输出；The non-scaling transformation blur device is connected to the instant statistics device and the bilateral filtering device respectively, and is used for receiving the bilateral filtering image, and when receiving the first control signal, the bilateral filtering image is used as the instant smooth image output;

数据辨识设备，位于婴儿床内，与所述无缩放变换模糊设备连接，用于接收所述即时平滑图像，基于婴儿脸部图像灰度阈值范围将所述即时平滑图像中的婴儿脸部识别出来以获得相应的目标区域，基于所述目标区域确定对应婴儿脸部的数个几何特征：欧拉孔数、圆度、角点数、凸凹度、光滑度、长径比、紧密度和主轴角度，并将所述数个几何特征组成特征向量；A data identification device, located in the crib, connected to the non-scaling transform blur device, for receiving the instant smoothed image, and identifying the infant face in the instant smoothed image based on the grayscale threshold range of the infant face image In order to obtain the corresponding target area, several geometric features corresponding to the baby's face are determined based on the target area: Euler hole number, roundness, corner number, convexity, smoothness, aspect ratio, compactness and main axis angle, and forming the several geometric features into a feature vector;

类型解析设备，与所述数据辨识设备连接，采用8输入4输出的单隐层BP神经网络，将所述数个几何特征作为输入层神经元，输出层为婴儿脸部类型，所述婴儿脸部类型包括疲倦脸部类型、饥饿脸部类型、痛苦脸部类型和缺水脸部类型；The type parsing device is connected to the data identification device, adopts a single hidden layer BP neural network with 8 inputs and 4 outputs, uses the several geometric features as input layer neurons, and the output layer is the baby face type, the baby face Face types include tired face, hungry face, distressed face and dehydrated face;

无线通信接口，与所述类型解析设备连接，用于将获取的婴儿脸部类型无线发送给婴儿家长的移动终端上；a wireless communication interface, connected with the type analysis device, for wirelessly sending the acquired baby face type to the mobile terminal of the baby's parent;

其中，所述无缩放变换模糊设备还用于在接收到所述第二控制信号时，对所述双边滤波图像执行无缩放变换模糊式的图像平滑处理，以获得并输出相应的即时平滑图像。Wherein, the non-scaling transform blurring device is further configured to perform non-scaling transform blurring image smoothing processing on the bilaterally filtered image when receiving the second control signal, so as to obtain and output a corresponding instant smoothed image.

接着，继续对本发明的基于定制数据处理的识别装置的具体结构进行进一步的说明。Next, the specific structure of the identification device based on customized data processing of the present invention will be further described.

所述基于定制数据处理的识别装置中：In the identification device based on customized data processing:

在所述曲线处理设备中，对接收到的畸变处理图像执行曲线处理，以获得对应的曲线处理图像包括：将接收到的畸变处理图像中的各个曲线分别对应的各个最大弧度都调节到预设弧度阈值之下。In the curve processing device, performing curve processing on the received distorted image to obtain the corresponding curve processed image includes: adjusting each maximum radian corresponding to each curve in the received distorted image to a preset value below the radian threshold.

所述基于定制数据处理的识别装置中还可以包括：The identification device based on customized data processing may also include:

SDRAM存储设备，用于预先存储所述纽扣采集设备设置拍摄方向的基准背景图以及左侧背景图和右侧背景图。The SDRAM storage device is used for pre-storing the reference background image, the left background image and the right background image of the button collecting device for setting the shooting direction.

所述基于定制数据处理的识别装置中：In the identification device based on customized data processing:

在所述SDRAM存储设备中，所述基准背景图为所述纽扣采集设备预先按照设置拍摄方向拍摄的无运动目标的图像，所述左侧背景图为所述纽扣采集设备预先按照设置拍摄方向左侧偏离后拍摄的无运动目标的图像，所述右侧背景图为所述纽扣采集设备预先按照设置拍摄方向右侧偏离后拍摄的无运动目标的图像。In the SDRAM storage device, the reference background image is an image without a moving target shot by the button collecting device in advance according to the set shooting direction, and the left background image is the button collecting device according to the preset shooting direction left in advance. An image without a moving target taken after the side is deviated, and the right background image is an image without a moving target captured by the button collecting device after being deviated to the right according to the set shooting direction in advance.

所述基于定制数据处理的识别装置中还可以包括：The identification device based on customized data processing may also include:

均值分析设备，分别与所述纽扣采集设备和所述SDRAM存储设备连接，用于接收所述床体环境图像、所述基准背景图、所述左侧背景图和所述右侧背景图，对所述床体环境图像的所有像素点的红色分量值进行求均值操作，以获取操作结果以作为目标分量值，对所述基准背景图的所有像素点的红色分量值进行求均值操作，以获取操作结果以作为背景分量值，对所述左侧背景图的所有像素点的红色分量值进行求均值操作，以获取操作结果以作为左侧分量值，对所述右侧背景图的所有像素点的红色分量值进行求均值操作，以获取操作结果以作为右侧分量值。The mean value analysis device is respectively connected with the button acquisition device and the SDRAM storage device, and is used for receiving the bed environment image, the reference background image, the left background image and the right background image, Perform an averaging operation on the red component values of all pixels of the bed environment image to obtain an operation result as a target component value, and perform an averaging operation on the red component values of all pixels in the reference background image to obtain The operation result is used as the background component value, and the average value operation is performed on the red component values of all the pixels of the left background image to obtain the operation result as the left component value, and all the pixels of the right background image are The value of the red component is averaged to obtain the result of the operation as the value of the right component.

所述基于定制数据处理的识别装置中还可以包括：The identification device based on customized data processing may also include:

均值匹配设备，与所述均值分析设备连接，用于接收所述目标分量值、所述背景分量值、所述左侧分量值以及所述右侧分量值。A mean value matching device, connected to the mean value analyzing device, is used for receiving the target component value, the background component value, the left component value and the right component value.

所述基于定制数据处理的识别装置中：In the identification device based on customized data processing:

所述均值匹配设备还用于计算所述目标分量值和所述背景分量值的差值，当所述差值超限时，发出方向改变信号，并将所述目标分量值分别与所述左侧分量值和所述右侧分量值匹配，当与所述左侧分量值的匹配度大于与所述右侧分量值的匹配度时，发出左向偏移信号，当与所述左侧分量值的匹配度小于等于与所述右侧分量值的匹配度时，发出右向偏移信号。The mean value matching device is also used to calculate the difference between the target component value and the background component value, when the difference exceeds the limit, a direction change signal is sent, and the target component value is respectively matched with the left side. The component value matches the right component value, and when the matching degree with the left component value is greater than the matching degree with the right component value, a left offset signal is sent, and when the matching degree with the left component value is greater than the matching degree with the right component value, a left offset signal is sent. When the matching degree of is less than or equal to the matching degree with the right component value, a right offset signal is sent.

所述基于定制数据处理的识别装置中还可以包括：The identification device based on customized data processing may also include:

驱动电机，分别与所述均值匹配设备和所述纽扣采集设备连接，用于在接收到所述左向偏移信号时，驱动所述纽扣采集设备向右侧偏转，还用于在接收到所述右向偏移信号时，驱动所述纽扣采集设备向左侧偏转。a driving motor, which is respectively connected with the mean value matching device and the button collecting device, and is used for driving the button collecting device to deflect to the right when receiving the left deviation signal, and is also used for receiving the When the rightward deviation signal is received, the button collecting device is driven to deflect to the left.

另外，SDRAM：Synchronous Dynamic Random Access Memory，同步动态随机存储器，同步是指内存工作需要同步时钟，内部的命令的发送与数据的传输都以他为基准；动态是指存储阵列需要不断的刷新来保证数据不丢失；随机是指数据不是线性依次存储，而是自由指定地址进行数据读写。SDR SDRAM的时钟频率就是数据存储的频率。SDRAM的工作电压为3.3V。In addition, SDRAM: Synchronous Dynamic Random Access Memory, synchronous dynamic random access memory, synchronization means that the memory work needs to synchronize the clock, and the transmission of internal commands and data transmission are based on him; dynamic means that the storage array needs to be constantly refreshed to ensure The data is not lost; random means that the data is not stored in a linear order, but freely specifies the address for data read and write. The clock frequency of SDR SDRAM is the frequency of data storage. The operating voltage of SDRAM is 3.3V.

在本说明书的描述中，参考术语“一个实施例”、“一些实施例”、“示例”、“具体示例”、或“一些示例”等的描述意指结合该实施例或示例描述的具体特征、结构、材料或者特点包含于本发明的至少一个实施例或示例中。In the description of this specification, description with reference to the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples", etc., mean specific features described in connection with the embodiment or example , structure, material or feature is included in at least one embodiment or example of the present invention.

此外，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括至少一个该特征。In addition, the terms "first" and "second" are only used for descriptive purposes, and should not be construed as indicating or implying relative importance or implying the number of indicated technical features. Thus, a feature delimited with "first", "second" may expressly or implicitly include at least one of that feature.

虽然本发明已以实施例揭示如上，但其并非用以限定本发明，任何所属技术领域的普通技术人员，在不脱离本发明的精神和范围内，应当可以做出适当的改动和同等替换。因此本发明的保护范围应当以本申请权利要求所界定的范围为准。Although the present invention has been disclosed above with embodiments, it is not intended to limit the present invention, and any person of ordinary skill in the technical field should be able to make appropriate changes and equivalent replacements without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention should be based on the scope defined by the claims of the present application.

Claims (8)

1. A recognition method based on customized data processing, the method comprising providing a recognition device based on customized data processing for introducing a neural network recognition mechanism to perform on-site recognition of infant facial types including a tired face type, a hungry face type, a painful face type and a water-deficient face type and wirelessly transmitting the recognition result to a parent's mobile terminal, the recognition device based on customized data processing comprising:

the button acquisition equipment is arranged above the crib and used for acquiring image data of the environment where the crib is located so as to obtain and output a corresponding bed body environment image;

the restoration processing equipment is positioned in the crib, is connected with the button acquisition equipment, and is used for receiving the bed body environment image and executing image restoration processing on the bed body environment image to obtain a real-time restoration image;

the real-time counting equipment is positioned in the crib, is connected with the restoration processing equipment, and is used for receiving the real-time restoration image, counting the data volume of the real-time restoration image and sending a first control signal when the counted data volume exceeds the limit;

the instant statistical equipment is also used for sending a second control signal when the statistical data volume does not exceed the limit;

in the instant statistics device, the performing statistics on the data size of the instant recovery image includes: identifying each redundant pixel point and each non-redundant pixel point in the instant restoration image, and calculating the data volume of the instant restoration image based on the product of the number of each non-redundant pixel point in the instant restoration image and the number of bits occupied by a single pixel point;

the distortion processing device is connected with the restoration processing device and used for receiving the instant restoration image and executing distortion correction processing on the instant restoration image so as to obtain and output a corresponding distortion processing image;

the curve processing device is connected with the distortion processing device and used for executing curve processing on the received distortion processing image to obtain a corresponding curve processing image;

the bilateral filtering equipment is connected with the curve processing equipment and is used for carrying out bilateral filtering processing on the received curve processing image so as to obtain and output a corresponding bilateral filtering image;

the non-scaling transformation fuzzy equipment is respectively connected with the instant statistic equipment and the bilateral filtering equipment and is used for receiving the bilateral filtering image and outputting the bilateral filtering image as an instant smooth image when receiving the first control signal;

the data identification device is positioned in the crib, is connected with the non-scaling transformation fuzzy device and is used for receiving the instant smooth image, identifying the infant face in the instant smooth image based on the gray level threshold range of the infant face image to obtain a corresponding target area, and determining a plurality of geometric features corresponding to the infant face based on the target area: the number of Euler holes, roundness, the number of corner points, convexity and concavity, smoothness, length-diameter ratio, compactness and main shaft angle, and forming a plurality of geometric features into a feature vector;

the type analysis device is connected with the data identification device, adopts a single hidden layer BP neural network with 8 input and 4 output, takes the plurality of geometric features as input layer neurons, and takes the output layer as an infant face type, wherein the infant face type comprises a tired face type, a hungry face type, a painful face type and a water-deficient face type;

the wireless communication interface is connected with the type analysis equipment and is used for wirelessly transmitting the acquired face type of the baby to a mobile terminal of a parent of the baby;

and the non-scaling transformation fuzzy equipment is further used for executing non-scaling transformation fuzzy type image smoothing processing on the bilateral filtering image when receiving the second control signal so as to obtain and output a corresponding instant smooth image.

2. The method of claim 1, wherein:

in the curve processing apparatus, performing curve processing on the received distortion-processed image to obtain a corresponding curve-processed image includes: and adjusting each maximum radian corresponding to each curve in the received distortion processing image to be below a preset radian threshold value.

3. The method of claim 2, wherein the apparatus further comprises:

and the SDRAM storage device is used for storing a reference background image, a left background image and a right background image of the shooting direction set by the button acquisition device in advance.

4. The method of claim 3, wherein:

in the SDRAM storage device, the reference background image is an image of a non-moving object shot by the button collecting device in advance according to a set shooting direction, the left background image is an image of a non-moving object shot by the button collecting device in advance according to a set shooting direction after left deviation, and the right background image is an image of a non-moving object shot by the button collecting device in advance according to a set shooting direction after right deviation.

5. The method of claim 4, wherein the apparatus further comprises:

and the mean value analysis equipment is respectively connected with the button acquisition equipment and the SDRAM storage equipment and is used for receiving the bed body environment image, the reference background image, the left background image and the right background image, averaging the red component values of all the pixel points of the bed body environment image to obtain an operation result as a target component value, averaging the red component values of all the pixel points of the reference background image to obtain an operation result as a background component value, averaging the red component values of all the pixel points of the left background image to obtain an operation result as a left component value, and averaging the red component values of all the pixel points of the right background image to obtain an operation result as a right component value.

6. The method of claim 5, wherein the apparatus further comprises:

a mean matching device connected with the mean analyzing device for receiving the object component value, the background component value, the left side component value, and the right side component value.

7. The method of claim 6, wherein:

the mean matching device is further configured to calculate a difference between the object component value and the background component value, issue a direction change signal when the difference is out of limit, and match the object component value with the left side component value and the right side component value, respectively, issue a left offset signal when a degree of matching with the left side component value is greater than a degree of matching with the right side component value, and issue a right offset signal when the degree of matching with the left side component value is less than or equal to the degree of matching with the right side component value.

8. The method of claim 7, wherein the apparatus further comprises:

and the driving motor is respectively connected with the mean value matching device and the button collecting device, and is used for driving the button collecting device to deflect towards the right side when receiving the leftward deviation signal and driving the button collecting device to deflect towards the left side when receiving the rightward deviation signal.

Images