Face recognition monitoring method based on edge computing

Abstract

The application provides a face recognition monitoring method based on edge computing, which segments target region images, and uses computing channels with different computing frequencies to recognize different sub-pictures after segmentation, so that suitable and fast recognition modes can be provided for different picture parts of the images, and accurate and fast recognition of all face pictures of the images is ensured.

Description

Technical Field

[0001] This invention relates to the technical field of face recognition, and in particular to a face recognition and monitoring method based on edge computing. Background Technology

[0002] Facial recognition has been widely used in the field of personnel identification. By capturing images of people's faces and identifying facial features, corresponding personal information can be obtained. Current facial recognition methods typically identify a single face within an image. When multiple faces are present in an image, each face needs to be identified individually. Furthermore, overlapping faces can lead to errors in correctly identifying facial features. Therefore, current facial recognition methods suffer from low efficiency and significant errors when multiple faces are present in the same image, failing to guarantee accurate and rapid identification of all faces. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a face recognition and monitoring method based on edge computing. It segments the target area image into at least one first sub-image and at least one second sub-image based on the personnel distribution information of the target area image. Using different computing channels of an edge computing terminal, the first and second sub-images are preprocessed and processed for face contour recognition to obtain facial feature information and determine the position of the person's face within the sub-image. The position information of all persons' faces in the same sub-image is mapped to the target area image. Based on the mapping result and the facial feature information, all persons in the target area image are identified and located. The method segments the target area image and uses computing channels with different computing frequencies to perform recognition processing on the different sub-images. This provides a suitable and fast recognition mode for different parts of the image, ensuring accurate and rapid recognition of all faces in the image.

[0004] This invention provides a face recognition and monitoring method based on edge computing, which includes the following steps:

[0005] Step S1: Take a picture of the target area to obtain an image of the target area; analyze and process the image of the target area to obtain personnel distribution information of the target area image; according to the personnel distribution information, divide the image of the target area into at least one first sub-image and at least one second sub-image.

[0006] Step S2: After generating first sub-screen queues and second sub-screen queues for all first sub-screens and all second sub-screens respectively, upload them to different computing channels of the edge computing terminal; then preprocess all sub-screens contained in the first sub-screen queue and the second sub-screen queue respectively.

[0007] Step S3: Perform facial contour recognition processing on the sub-screen to obtain facial feature information of the person in the sub-screen; determine the position information of the corresponding person's face in the sub-screen based on the facial feature information; map the position information of the faces of all persons in the same sub-screen to the target area image.

[0008] Step S4: Based on the mapping result and the facial contour features of the person, identify and locate all persons present in the target area image.

[0009] Furthermore, in step S1, the target area is photographed to obtain an image of the target area; the target area image is analyzed and processed to obtain the personnel distribution information of the target area image, specifically including:

[0010] The target area is scanned and photographed to obtain a panoramic image of the target area; the body contour feature information of all people is extracted from the panoramic image of the target area, and the body center of gravity position information of each person is determined based on the body contour feature information;

[0011] Then, based on the individual body center of gravity position information of each person, the relative distance information between each person is determined, which is used as the personnel distribution status information.

[0012] Furthermore, in step S1, dividing the target area image into at least one first sub-image and at least one second sub-image based on the personnel distribution status information specifically includes:

[0013] Based on the relative distance information between all personnel, all personnel present in the target area image are divided into at least a first personnel set and at least a second personnel set; wherein, the average distance between all personnel in the first personnel set is less than the average distance between all personnel in the second personnel set.

[0014] The portion of the image occupied by each first group of personnel and each second group of personnel in the target area is respectively designated as the first sub-image and the second sub-image; wherein, all first sub-images and all second sub-images do not overlap with each other.

[0015] Furthermore, in step S2, after generating first sub-screen queues and second sub-screen queues for all first sub-screens and all second sub-screens respectively, and uploading them to different computing channels of the edge computing terminal, specifically includes:

[0016] Obtain the contrast between the person and the background in each of the first and second sub-screens; arrange all the first and second sub-screens into a first sub-screen queue and a second sub-screen queue according to the order of the contrast from largest to smallest.

[0017] The first sub-screen queue and the second sub-screen queue are uploaded to the first computing channel and the second computing channel of the edge computing terminal, respectively; wherein the computing frequency of the first computing channel is greater than the computing frequency of the second computing channel.

[0018] Furthermore, in step S2, the preprocessing of all sub-pictures contained in the first sub-picture queue and the second sub-picture queue specifically includes:

[0019] The first calculation channel and the second calculation channel respectively perform noise reduction filtering and pixel sharpening processing on all sub-pictures contained in the first sub-picture queue and the second sub-picture queue.

[0020] Further, in step S3, facial contour recognition processing is performed on the sub-screen to obtain the facial feature information of the person contained in the sub-screen; determining the position information of the corresponding person's face in the sub-screen based on the facial feature information specifically includes:

[0021] Face contour recognition processing is performed on each sub-screen contained in the first sub-screen queue and the second sub-screen queue to obtain the facial edge contour feature information and facial feature contour feature information of each person in the sub-screen.

[0022] Based on the facial edge contour feature information, the position information of the geometric center of the corresponding person's face in the sub-screen is determined.

[0023] Furthermore, in step S3, mapping the positional information of the faces of all persons existing in the same sub-frame to the target area image specifically includes:

[0024] Based on the coordinate transformation relationship between the sub-screen and the target area image, and the position information of the geometric center of the face of all people in the sub-screen in the sub-screen, the position information of the geometric center of the face of all people in the sub-screen is mapped to the target area image.

[0025] Furthermore, in step S4, the identification and location identification of all persons present in the target area image based on the mapping result and the facial contour feature information of the persons specifically includes:

[0026] Based on the mapping results, the position information of the geometric center of each person's face on the entire image of the target area is determined;

[0027] Based on the facial contour features, the identity information of each person is identified from a preset identity information database.

[0028] Compared to existing technologies, this edge computing-based facial recognition monitoring method segments the target area image into at least one first sub-image and at least one second sub-image based on the personnel distribution information of the target area image. It then uses different computing channels of the edge computing terminal to preprocess the first and second sub-images and perform facial contour recognition processing to obtain facial feature information and determine the position of the person's face within the sub-image. The position information of all persons' faces in the same sub-image is mapped to the target area image. Based on the mapping results and the facial feature information, all persons in the target area image are identified and located. The method segments the target area image and uses computing channels with different computing frequencies to process the segmented sub-images, thus providing a suitable and fast recognition mode for different parts of the image and ensuring accurate and rapid recognition of all faces in the image.

[0029] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0030] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 This is a flowchart illustrating the face recognition and monitoring method based on edge computing provided by the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] See Figure 1 This is a flowchart illustrating the face recognition and monitoring method based on edge computing provided in an embodiment of the present invention. The face recognition and monitoring method based on edge computing includes the following steps:

[0035] Step S1: Take a picture of the target area to obtain an image of the target area; analyze and process the image of the target area to obtain the personnel distribution status information of the target area image; based on the personnel distribution status information, divide the image of the target area into at least one first sub-image and at least one second sub-image.

[0036] Step S2: After generating first sub-screen queues and second sub-screen queues for all first sub-screens and all second sub-screens respectively, upload them to different computing channels of the edge computing terminal; then preprocess all sub-screens contained in each of the first sub-screen queues and the second sub-screen queues.

[0037] Step S3: Perform facial contour recognition processing on the sub-screen to obtain the facial contour feature information of the person in the sub-screen; determine the position information of the corresponding person's face in the sub-screen based on the facial contour feature information; map the position information of the faces of all persons in the same sub-screen to the target area image.

[0038] Step S4: Based on the mapping result and the facial contour features of the person, identify and locate all persons present in the target area image.

[0039] The beneficial effects of the above technical solution are as follows: This edge computing-based facial recognition monitoring method divides the target area image into at least one first sub-image and at least one second sub-image based on the personnel distribution information of the target area image; it uses different computing channels of the edge computing terminal to preprocess the first and second sub-images and perform facial contour recognition processing to obtain the facial feature information of the personnel and determine the position information of the personnel's faces in the sub-images; it maps the position information of the faces of all personnel in the same sub-image to the target area image; and based on the mapping results and the facial feature information of the personnel, it identifies and locates all personnel in the target area image. It segments the target area image and uses computing channels with different computing frequencies to perform recognition processing on the different sub-images after segmentation. This can provide a suitable and fast recognition mode for different parts of the image, ensuring accurate and fast recognition of all facial images.

[0040] Preferably, in step S1, the target area is photographed to obtain an image of the target area; the image of the target area is analyzed and processed to obtain the personnel distribution information of the target area, specifically including:

[0041] The target area is scanned and photographed to obtain a panoramic image of the target area; the body contour feature information of all people is extracted from the panoramic image of the target area, and the body center of gravity position information of each person is determined based on the body contour feature information;

[0042] Then, based on the center of gravity position information of each person, the relative distance information between each person is determined, which is used as the distribution status information of the person.

[0043] The beneficial effects of the above technical solution are as follows: By using scanning and imaging, all personnel in the target area can be completely captured in the panoramic image. In the panoramic image, each person's position is not uniform, resulting in groups of people clustered together in close proximity. Based on the body contour features of each person in the panoramic image, the position of each person's center of gravity is determined. Then, based on the center of gravity positions of all personnel, the relative distances between them are determined, thereby quantitatively identifying the distribution of all personnel in the panoramic image.

[0044] Preferably, in step S1, dividing the target area image into at least one first sub-image and at least one second sub-image based on the personnel distribution information specifically includes:

[0045] Based on the relative distance information between all personnel, all personnel present in the target area image are divided into at least a first personnel set and at least a second personnel set; wherein, the average distance between all personnel in the first personnel set is less than the average distance between all personnel in the second personnel set.

[0046] The portion of the image occupied by each first group of personnel and each second group of personnel in the target area is designated as the first sub-image and the second sub-image, respectively; wherein, all first sub-images and all second sub-images do not overlap with each other.

[0047] The beneficial effects of the above technical solution are as follows: In practical work, based on the relative distance information between all personnel and using a preset distance threshold as a benchmark, if the average distance between all personnel in a crowd formed by several personnel gathering is less than the preset distance threshold, then the corresponding crowd is determined as the first personnel set; if the average distance between all personnel in a crowd formed by several personnel gathering is greater than the preset distance threshold, then the corresponding crowd is determined as the second personnel set; then the boundary lines corresponding to each first personnel set and each second personnel set in the panoramic image of the target area are determined, and the panoramic image of the target area is segmented based on the boundary lines to obtain the corresponding first sub-image and second sub-image.

[0048] Preferably, in step S2, after generating first sub-screen queues and second sub-screen queues for all first sub-screens and all second sub-screens respectively, uploading them to different computing channels of the edge computing terminal specifically includes:

[0049] Get the contrast between the person and the background in each of the first and second sub-screens; arrange all the first and second sub-screens into a first sub-screen queue and a second sub-screen queue according to the order of the contrast from largest to smallest.

[0050] The first sub-screen queue and the second sub-screen queue are uploaded to the first computing channel and the second computing channel of the edge computing terminal, respectively; wherein the computing frequency of the first computing channel is greater than the computing frequency of the second computing channel.

[0051] The beneficial effects of the above technical solution are as follows: By using the contrast between the person and background in a sub-frame as a benchmark, all first sub-frames and all second sub-frames are sorted to form a first sub-frame queue and a second sub-frame queue. This facilitates the orderly processing of the received sub-frame queues by the corresponding subsequent computing channels. Furthermore, the distribution of people in each sub-frame in the first sub-frame queue is relatively dense, while the distribution in each sub-frame in the second sub-frame queue is relatively sparse. This means that face recognition in the first sub-frame queue is more difficult than in the second sub-frame queue. Therefore, using the first computing channel with a higher computational frequency to process the first sub-frame queue improves the efficiency and accuracy of face recognition in the first sub-frame queue.

[0052] Preferably, in step S2, the preprocessing of all sub-pictures contained in the first sub-picture queue and the second sub-picture queue specifically includes:

[0053] The first calculation channel and the second calculation channel respectively perform noise reduction filtering and pixel sharpening processing on all sub-pictures contained in the first sub-picture queue and the second sub-picture queue.

[0054] The beneficial effects of the above technical solution are as follows: through the above method, it is possible to effectively reduce noise and enhance pixels in all sub-pictures contained in the first sub-picture queue and the second sub-picture queue, which facilitates the accurate identification of facial feature elements in the sub-pictures.

[0055] Preferably, in step S3, facial contour recognition processing is performed on the sub-screen to obtain the facial feature information of the person contained in the sub-screen; determining the position information of the corresponding person's face in the sub-screen based on the facial feature information specifically includes:

[0056] Face contour recognition processing is performed on each sub-screen contained in the first sub-screen queue and the second sub-screen queue to obtain the facial edge contour feature information and facial feature contour feature information of each person in the sub-screen.

[0057] Based on the facial edge contour features, the position of the geometric center of the corresponding person's face in the sub-image is determined.

[0058] The beneficial effects of the above technical solution are as follows: by using the facial edge contour feature information of the person, the location of the person's face on the sub-screen can be simplified and marked. At the same time, the facial feature information of each person can be obtained, and the identity of the person can be uniquely marked.

[0059] Preferably, in step S3, mapping the positional information of the faces of all persons existing in the same sub-frame to the target area image specifically includes:

[0060] Based on the coordinate transformation relationship between the sub-screen and the target area image, and the position information of the geometric center of the face of all people in the sub-screen, the position information of the geometric center of the face of all people in the sub-screen is mapped to the target area image.

[0061] The beneficial effects of the above technical solution are as follows: each sub-screen is segmented from the target area image, so there is a corresponding spatial coordinate transformation relationship between each sub-screen and the target area image. At this time, by combining the position information of the geometric center of the face of all people in the sub-screen in the sub-screen, the geometric center of the face of each person in the sub-screen can be transformed into the target area image, thereby calibrating the position of the geometric center of the face of each person in the target area image.

[0062] Preferably, in step S4, the identification and location identification of all persons present in the target area image based on the mapping result and the facial contour feature information of the person specifically includes:

[0063] Based on the mapping result, the position information of the geometric center of each person's face on the entire image of the target area is determined;

[0064] Based on the facial contour features, the identity information of each person is identified from a pre-set identity information database.

[0065] The beneficial effects of the above technical solution are as follows: through the above method, the location information and identity information of each person can be marked one by one in the target area image, which facilitates the dual identification of the location and identity of all persons in the target area.

[0066] As can be seen from the above embodiments, this edge computing-based face recognition monitoring method divides the target area image into at least one first sub-image and at least one second sub-image based on the personnel distribution information of the target area image; it uses different computing channels of the edge computing terminal to preprocess the first sub-image and the second sub-image respectively to perform face contour recognition processing, obtain the facial feature information of the personnel, and determine the position information of the personnel's face in the sub-image; it maps the position information of the faces of all personnel in the same sub-image to the target area image; and then, based on the mapping result and the facial feature information of the personnel, it identifies and locates all personnel in the target area image. This method segments the target area image and uses computing channels with different computing frequencies to perform recognition processing on the different sub-images after segmentation. This provides a suitable and fast recognition mode for different parts of the image, ensuring accurate and fast recognition of all faces in the image.

[0067] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A face recognition and monitoring method based on edge computing, characterized in that, It includes the following steps: Step S1: Take a picture of the target area to obtain an image of the target area; analyze and process the image of the target area to obtain the personnel distribution information of the target area, including: The target area is scanned and photographed to obtain a panoramic image of the target area; the body contour feature information of all people is extracted from the panoramic image of the target area; the body center of gravity position information of each person is determined based on the body contour feature information; and the relative distance information between all people is determined based on the body center of gravity position information of each person, which is used as the personnel distribution status information. Based on the personnel distribution information, the target area image is divided into at least one first sub-image and at least one second sub-image. Step S2: After generating first sub-screen queues and second sub-screen queues for all first sub-screens and all second sub-screens respectively, upload them to different computing channels of the edge computing terminal; then preprocess all sub-screens contained in the first sub-screen queue and the second sub-screen queue respectively. Step S3: Perform facial contour recognition processing on the sub-screen to obtain facial feature information of the person in the sub-screen; determine the position information of the corresponding person's face in the sub-screen based on the facial feature information; map the position information of the faces of all persons in the same sub-screen to the target area image. Step S4: Based on the mapping result and the facial contour features of the person, identify and locate all persons present in the target area image.

2. The face recognition and monitoring method based on edge computing as described in claim 1, characterized in that: In step S1, dividing the target area image into at least one first sub-image and at least one second sub-image based on the personnel distribution status information specifically includes: Based on the relative distance information between all personnel, all personnel present in the target area image are divided into at least a first personnel set and at least a second personnel set; wherein, the average distance between all personnel in the first personnel set is less than the average distance between all personnel in the second personnel set. The portion of the image occupied by each first group of personnel and each second group of personnel in the target area is respectively designated as the first sub-image and the second sub-image; wherein, all first sub-images and all second sub-images do not overlap with each other.

3. The face recognition and monitoring method based on edge computing as described in claim 2, characterized in that: In step S2, after generating first sub-screen queues and second sub-screen queues for all first sub-screens and all second sub-screens respectively, they are uploaded to different computing channels of the edge computing terminal, specifically including: Obtain the contrast between the person and the background in each of the first and second sub-screens; arrange all the first and second sub-screens into a first sub-screen queue and a second sub-screen queue according to the order of the contrast from largest to smallest. The first sub-screen queue and the second sub-screen queue are uploaded to the first computing channel and the second computing channel of the edge computing terminal, respectively; wherein the computing frequency of the first computing channel is greater than the computing frequency of the second computing channel.

4. The face recognition and monitoring method based on edge computing as described in claim 3, characterized in that: In step S2, the preprocessing of all sub-pictures contained in the first sub-picture queue and the second sub-picture queue specifically includes: The first calculation channel and the second calculation channel respectively perform noise reduction filtering and pixel sharpening processing on all sub-pictures contained in the first sub-picture queue and the second sub-picture queue.

5. The face recognition and monitoring method based on edge computing as described in claim 4, characterized in that: In step S3, facial contour recognition processing is performed on the sub-screen to obtain the facial contour feature information of the person contained in the sub-screen. Determining the position information of the corresponding person's face in the sub-screen based on the facial contour features of the person specifically includes: Face contour recognition processing is performed on each sub-screen contained in the first sub-screen queue and the second sub-screen queue to obtain the facial edge contour feature information and facial feature contour feature information of each person in the sub-screen. Based on the facial edge contour feature information, the position information of the geometric center of the corresponding person's face in the sub-screen is determined.

6. The face recognition and monitoring method based on edge computing as described in claim 5, characterized in that: In step S3, mapping the positional information of the faces of all persons in the same sub-frame to the target area image specifically includes: Based on the coordinate transformation relationship between the sub-screen and the target area image, and the position information of the geometric center of the face of all people in the sub-screen in the sub-screen, the position information of the geometric center of the face of all people in the sub-screen is mapped to the target area image.

7. The face recognition and monitoring method based on edge computing as described in claim 6, characterized in that: In step S4, based on the mapping result and the facial contour feature information of the individuals, the identification and location identification of all individuals present in the target area image specifically includes: Based on the mapping results, the position information of the geometric center of each person's face on the entire image of the target area is determined; Based on the facial contour features, the identity information of each person is identified from a preset identity information database.

Images