Dynamic capture analysis algorithm based on video images

Abstract

The present invention discloses a dynamic capture analysis algorithm based on video images, relates to the technical field of video image dynamic capture analysis, solves the technical problem in the prior art that the action comparison of the corresponding subject cannot be accurately determined during dynamic capture, resulting in low capture efficiency, and determines whether the quality of the real-time video image is qualified, thereby ensuring the qualification of the dynamic capture and improving the work efficiency of the dynamic capture; regularizes areas with the same features, thereby reducing the work intensity of the video image analysis, and simultaneously prevents repeated analysis of the same feature, resulting in unnecessary cost increase, and simultaneously increasing the risk of dynamic analysis errors; determines whether the picture of the current frame is in a static state, optimizes the static part, reduces the resolution of the static part, reduces the work intensity of the dynamic capture, and improves the work efficiency of the video image dynamic capture; and analyzes the internal dynamic action of the real-time video image to determine whether the current dynamic action is normal.

Description

Technical Field

[0001] The present invention relates to the technical field of video image dynamic capture and analysis, and in particular to a video image-based dynamic capture and analysis algorithm. Background Art

[0002] As we all know, humans obtain, process and understand visual information through their eyes and brains. Image visual information is one of the main sources of information that humans obtain from the objective world. Video images are actually composed of a series of connected static images. They are a vivid description of objective things and an intuitive and specific form of information expression. After entering the 1990s, with the rapid development of computer networks, communications and microelectronics technology, video surveillance has been increasingly valued and studied by people for its intuitive, convenient and rich information content. The continuous improvement of requirements for personal and property safety has promoted people's demand for monitoring systems. In the past, the care of some important places such as warehouses and banks was done by people. But in modern society, there are more and more places that need to be monitored, and the requirements for monitoring are getting higher and higher (long-term uninterrupted monitoring is required to prevent illegal intrusion, etc.), so the importance and practicality of modern monitoring systems are reflected-it can complete long-term uninterrupted monitoring of important places (banks, supermarkets, warehouses, museums), and send out alarm signals in time after illegal intrusion.

[0003] However, in the prior art, when video images are dynamically captured, the investment cost of video image analysis cannot be reduced through reasonable processing, resulting in unnecessary waste of costs; at the same time, the images cannot be properly planned, resulting in excessively high costs for image preservation and transmission during the dynamic capture process; in addition, the corresponding subject's action comparison cannot be accurately determined during dynamic capture, and the dynamic capture work efficiency is low.

[0004] In view of the above technical defects, a solution is now proposed. Summary of the invention

[0005] The purpose of the present invention is to solve the above-mentioned problems and to propose a dynamic capture analysis algorithm based on video images, which analyzes the corresponding real-time video images to determine whether the quality of the real-time video images is qualified, thereby ensuring the qualification of dynamic capture and improving the work efficiency of dynamic capture; reasonably divides each area and regularizes the areas with the same characteristics, which can reduce the workload of video image analysis and prevent repeated analysis of the same feature, resulting in unnecessary cost increase and increased risk of dynamic analysis errors; determines whether the picture of the current frame is static, optimizes the static part, reduces the resolution of the static part, reduces the workload of dynamic capture, and improves the work efficiency of video image dynamic capture; and analyzes the internal dynamic action of the real-time video image to determine whether the current dynamic action is normal.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] Based on the dynamic capture analysis algorithm of video images, the specific dynamic capture analysis steps are as follows:

[0008] S1: Image analysis: The video is acquired through the real-time monitoring device and marked as a real-time video image. After the real-time video image is acquired, the corresponding real-time video image is analyzed to determine whether the quality of the real-time video image is qualified;

[0009] S2: Region segmentation: perform region segmentation on the video image that has passed the image analysis, divide the video image into i frames, where i is a natural number greater than 1, and collect pixels in each frame. The distribution of the pixels in the frame is obtained through the collected pixels, and feature comparison is performed on the region of the frame where the pixel is located. Region segmentation is performed based on the feature comparison, and each frame is divided into o sub-regions.

[0010] S3: Frame reduction optimization, analyzing the divided frame images of the video image to determine whether the image of the current frame is static, and optimizing the static part to reduce the resolution of the static part;

[0011] S4: Dynamic capture, analyzing each dynamic part in the frame image that has completed static optimization, collecting the corresponding dynamic part of each frame image, playing each frame image in a dynamic form, analyzing the corresponding dynamic part of each frame image, acquiring the dynamic action in the real-time video image, and analyzing the dynamic action in the real-time video image to determine whether the current dynamic action is normal.

[0012] As a preferred embodiment of the present invention, the specific steps of image analysis in step S1 are as follows:

[0013] S11: The video is acquired through a real-time monitoring device, marked as a real-time video image, and the quality of the video image is analyzed;

[0014] S12: Collect the average number of frames per second of the real-time video image and the resolution of the corresponding frame image, and mark the average number of frames per second of the real-time video image and the resolution of the corresponding frame image as ZS and FB respectively; by formula X = (ZS × a1 + FB × a2) a1+a2 The quality analysis coefficient X of the real-time video image is obtained, wherein a1 and a2 are both preset proportional coefficients, and a1 is greater than a2 and greater than 0;

[0015] S13: Compare the quality analysis coefficient X of the real-time video image with the quality analysis coefficient threshold:

[0016] If the quality analysis coefficient X of the real-time video image exceeds the quality analysis coefficient threshold, the corresponding real-time video image quality analysis is judged to be qualified, a quality analysis qualified signal is generated and the video image is marked as a qualified video image, and the process proceeds to step S2; if the quality analysis coefficient X of the real-time video image does not exceed the quality analysis coefficient threshold, the corresponding real-time video image quality analysis is judged to be unqualified, a quality analysis unqualified signal is generated and the video image is marked as an unqualified video image, and no motion capture is performed.

[0017] As a preferred embodiment of the present invention, the specific steps of region segmentation in step S2 are as follows:

[0018] Step S21: segmenting the qualified video image into regions, analyzing the frame images divided into the video image, and performing feature comparison based on the analysis of the pixels in each frame image;

[0019] Step S22: setting a pixel number threshold, and dividing the region in the frame image formed by the pixel number threshold into o sub-regions, and the sub-regions corresponding to the pixel number threshold have the same features, and the sub-region features are represented by the brightness and saturation of the corresponding sub-regions in the frame image;

[0020] Step S23: If the corresponding values ​​of the data in the features in adjacent sub-regions are all within the corresponding data value threshold range, then the adjacent sub-regions are determined to be marked as sub-regions with the same features; if the corresponding values ​​of the data in the features in the adjacent sub-regions are not all within the corresponding data value threshold range, then the adjacent sub-regions are determined to be marked as sub-regions with different features; the corresponding sub-regions with the same features in the frame image are regularized, and the regularized sub-regions and sub-regions with different features are uniformly marked as analysis sub-regions, and the features in the same analysis sub-region are consistent, and the features in adjacent analysis sub-regions are inconsistent.

[0021] As a preferred implementation of the present invention, the specific steps of frame reduction optimization in step S3 are as follows:

[0022] Step S31: Arrange the frame pictures corresponding to the video image in the order in which the video image is divided, analyze the analysis sub-area in each frame picture, obtain the number of subjects in the analysis sub-area and the pixel position of the corresponding subject, and analyze the number of subjects in the corresponding analysis sub-area in adjacent frame pictures and the pixel position of the corresponding subject;

[0023] Step S32: If there is a numerical fluctuation in the number of subjects in the corresponding analysis sub-region in the adjacent frame pictures, or there is a pixel position change in the pixel position of the corresponding subject in the corresponding analysis sub-region in the adjacent frame pictures, it is determined that there is a dynamic operation in the current analysis sub-region, and the corresponding analysis sub-region is marked as a dynamic sub-region; if there is no numerical fluctuation in the number of subjects in the corresponding analysis sub-region in the adjacent frame pictures, and there is no pixel position change in the pixel position of the corresponding subject in the corresponding analysis sub-region in the adjacent frame pictures, it is determined that there is no dynamic operation in the current analysis sub-region, and the corresponding analysis sub-region is marked as a static sub-region;

[0024] Step S33: Control the resolution of the static sub-region in the corresponding frame image, and the resolution control standard is that the resolution of the static sub-region meets the normal image resolution threshold, and the resolution of the static sub-region is smaller than the resolution of the dynamic sub-region.

[0025] As a preferred embodiment of the present invention, the specific steps of dynamic capture in step S4 are as follows:

[0026] Step S41: obtaining a frame image with a dynamic sub-region, analyzing the corresponding frame image, playing the frame image with the dynamic sub-region in a dynamic form, obtaining a position change trajectory of the corresponding subject in the dynamic sub-region through the change of the position of the corresponding subject pixel point in the dynamic sub-region, and marking the corresponding position change trajectory as a real-time subject action;

[0027] Step S42: marking the corresponding dynamic sub-region as an action analysis region, limiting the action of the corresponding subject in the action analysis region, and marking the action that completes the limitation as a preset subject action;

[0028] Step S43: Compare the real-time subject action of the corresponding subject in the action analysis area with the preset subject action. If the real-time subject action is consistent with the preset subject action, the real-time action of the corresponding subject is determined to be abnormal, an abnormal action signal is generated and the abnormal action signal is sent to the mobile phone terminal of the motion capture executor; if the real-time subject action is not consistent with the preset subject action, the real-time action of the corresponding subject is determined to be normal, a normal action signal is generated and the normal action signal is sent to the mobile phone terminal of the motion capture executor.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] In the present invention, the corresponding real-time video image is analyzed to determine whether the quality of the real-time video image is qualified, thereby ensuring the qualification of dynamic capture and improving the work efficiency of dynamic capture; each area is reasonably divided and the areas with the same characteristics are regularized, which can reduce the work intensity of video image analysis and prevent the same feature from being repeatedly analyzed, resulting in unnecessary cost increase and increased risk of dynamic analysis error; it is determined whether the picture of the current frame is static, and the static part is optimized, the resolution of the static part is reduced, the work intensity of dynamic capture is reduced, and the work efficiency of video image dynamic capture is improved; and the internal dynamic action of the real-time video image is analyzed to determine whether the current dynamic action is normal; it is possible to accurately determine whether the subject action in the corresponding area is normal through the video image, thereby improving the monitoring efficiency of the video image and enhancing the qualification and work efficiency of video image dynamic capture. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0032] Figure 1 The present invention is a method flow chart of a dynamic capture analysis algorithm based on video images. DETAILED DESCRIPTION

[0033] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] See also Figure 1 As shown in the figure, the dynamic capture analysis algorithm based on video images, the specific dynamic capture analysis steps are as follows:

[0035] S1: Image analysis: The video is acquired through the real-time monitoring device and marked as a real-time video image. After the real-time video image is acquired, the corresponding real-time video image is analyzed to determine whether the quality of the real-time video image is qualified, thereby ensuring the qualification of dynamic capture and improving the work efficiency of dynamic capture;

[0036] S2: Region segmentation: perform region segmentation on the video image that has passed the image analysis, divide the video image into i frames, where i is a natural number greater than 1, and collect pixel points in each frame, obtain the distribution of the pixel points of the frame through the collected pixel points, and compare the features of the region of the frame where the pixel points are located. Perform region segmentation based on the feature comparison, and divide each frame into o sub-regions; reasonably segment each region and regularize the regions with the same features, which can reduce the workload of video image analysis and prevent repeated analysis of the same feature, resulting in unnecessary cost increase and increased risk of dynamic analysis errors;

[0037] S3: Frame reduction optimization, analyze the frame pictures after the video image is divided, determine whether the picture of the current frame is static, and optimize the static part, reduce the resolution of the static part, reduce the workload of dynamic capture, and improve the work efficiency of video image dynamic capture;

[0038] S4: Dynamic capture, analyzing each dynamic part in the frame image that has completed static optimization, collecting the corresponding dynamic part of each frame image, playing each frame image in a dynamic form, analyzing the corresponding dynamic part of each frame image, obtaining the dynamic action in the real-time video image, and analyzing the dynamic action in the real-time video image to determine whether the current dynamic action is normal; it can accurately determine whether the subject action in the corresponding area is normal through the video image, thereby improving the monitoring efficiency of the video image, and enhancing the qualification and work efficiency of the video image dynamic capture;

[0039] The specific steps of image analysis in step S1 are as follows:

[0040] S11: The video is acquired through a real-time monitoring device, marked as a real-time video image, and the quality of the video image is analyzed;

[0041] S12: Collect the average number of frames per second of the real-time video image and the resolution of the corresponding frame image, and mark the average number of frames per second of the real-time video image and the resolution of the corresponding frame image as ZS and FB respectively; by formula X = (ZS × a1 + FB × a2) a1+a2 The quality analysis coefficient X of the real-time video image is obtained, wherein a1 and a2 are both preset proportional coefficients, and a1 is greater than a2 and greater than 0;

[0042] S13: Compare the quality analysis coefficient X of the real-time video image with the quality analysis coefficient threshold:

[0043] If the quality analysis coefficient X of the real-time video image exceeds the quality analysis coefficient threshold, the quality analysis of the corresponding real-time video image is determined to be qualified, a quality analysis qualified signal is generated, and the video image is marked as a qualified video image, and the process proceeds to step S2; if the quality analysis coefficient X of the real-time video image does not exceed the quality analysis coefficient threshold, the quality analysis of the corresponding real-time video image is determined to be unqualified, a quality analysis unqualified signal is generated, and the video image is marked as an unqualified video image, and no dynamic capture is performed;

[0044] The specific steps of region segmentation in step S2 are as follows:

[0045] Step S21: segmenting the qualified video image into regions, analyzing the frame images divided into the video image, and performing feature comparison based on the analysis of the pixels in each frame image;

[0046] Step S22: setting a pixel number threshold, and dividing the region in the frame image formed by the pixel number threshold into o sub-regions, and the sub-regions corresponding to the pixel number threshold have the same features, and the sub-region features are represented by the brightness and saturation of the corresponding sub-regions in the frame image;

[0047] Step S23: if the corresponding values ​​of the data in the features of the adjacent sub-regions are all within the corresponding data value threshold range, the adjacent sub-regions are determined to be marked as the same-feature sub-regions; if the corresponding values ​​of the data in the features of the adjacent sub-regions are not all within the corresponding data value threshold range, the adjacent sub-regions are determined to be marked as non-same-feature sub-regions; the corresponding same-feature sub-regions in the frame image are regularized, and the regularized sub-regions and non-same-feature sub-regions are uniformly marked as analysis sub-regions, and the features in the same analysis sub-region are consistent, and the features in adjacent analysis sub-regions are inconsistent;

[0048] The specific steps of frame reduction optimization in step S3 are as follows:

[0049] Step S31: Arrange the frame pictures corresponding to the video image in the order in which the video image is divided, analyze the analysis sub-areas in each frame picture, obtain the number of subjects in the analysis sub-area and the pixel positions of the corresponding subjects, and analyze the number of subjects in the corresponding analysis sub-areas in adjacent frame pictures and the pixel positions of the corresponding subjects; in this application, the subjects in the analysis area are represented as objects existing in the video image, such as: if the video image is a desktop image, movable objects such as cups on the desktop can be set as subjects as a reference standard;

[0050] Step S32: If there is a numerical fluctuation in the number of subjects in the corresponding analysis sub-region in the adjacent frame pictures, or there is a pixel position change in the pixel position of the corresponding subject in the corresponding analysis sub-region in the adjacent frame pictures, it is determined that there is a dynamic operation in the current analysis sub-region, and the corresponding analysis sub-region is marked as a dynamic sub-region; if there is no numerical fluctuation in the number of subjects in the corresponding analysis sub-region in the adjacent frame pictures, and there is no pixel position change in the pixel position of the corresponding subject in the corresponding analysis sub-region in the adjacent frame pictures, it is determined that there is no dynamic operation in the current analysis sub-region, and the corresponding analysis sub-region is marked as a static sub-region;

[0051] Step S33: controlling the resolution of the static sub-region in the corresponding frame image, and the resolution control standard is that the resolution of the static sub-region meets the normal image resolution threshold, and the resolution of the static sub-region is smaller than the resolution of the dynamic sub-region;

[0052] The specific steps of dynamic capture in step S4 are as follows:

[0053] Step S41: Obtain a frame image in which a dynamic sub-area exists, and analyze the corresponding frame image, play the frame image in which the dynamic sub-area exists in a dynamic form, obtain the position change trajectory of the corresponding subject in the dynamic sub-area through the change of the position of the corresponding subject pixel point in the dynamic sub-area, and mark the corresponding position change trajectory as a real-time subject action; in this application, the subject action dynamic capture is used in the monitoring field, wherein the corresponding subject is a person, and the corresponding real-time position of the subject person in each frame image in the corresponding monitoring video image is dynamically captured, and the position change trajectory of the subject person in the dynamic sub-area can be used to determine whether the subject person stays or leaves in the current area, and whether the current area is used as an arrival area or a departure area;

[0054] Step S42: marking the corresponding dynamic sub-area as an action analysis area, limiting the action of the corresponding subject in the action analysis area, and marking the action that completes the limitation as a preset subject action; in this application, the subject is a person as a reference, and the action limitation is expressed as limiting the subject's staying action or wandering action in the action analysis area;

[0055] Step S43: Compare the real-time subject action of the corresponding subject in the action analysis area with the preset subject action. If the real-time subject action is consistent with the preset subject action, the real-time action of the corresponding subject is determined to be abnormal, an abnormal action signal is generated and the abnormal action signal is sent to the mobile phone terminal of the motion capture executor; if the real-time subject action is not consistent with the preset subject action, the real-time action of the corresponding subject is determined to be normal, a normal action signal is generated and the normal action signal is sent to the mobile phone terminal of the motion capture executor.

[0056] The above formulas are obtained by collecting a large amount of data and performing software simulation, and a formula close to the actual value is selected. The coefficients in the formula are set by technicians in this field according to actual conditions;

[0057] When the present invention is used, image analysis is performed, in which a video is acquired through a real-time monitoring device and marked as a real-time video image. After the acquisition of the real-time video image is completed, the corresponding real-time video image is analyzed to determine whether the quality of the real-time video image is qualified; region segmentation is performed, in which a video image that has passed the image analysis is region segmented, the video image is divided into i frame images, i is a natural number greater than 1, and pixel points in each frame image are collected, the distribution of the pixel points of the frame image is obtained through the collected pixel points, and the region of the frame image where the pixel points are located is compared with the characteristics, and region segmentation is performed according to the feature comparison, and each frame image is divided into o sub-regions; frame reduction optimization is performed, in which the frame image after the video image is divided is analyzed to determine whether the image of the current frame is in a static state, and the static part is optimized to reduce the resolution of the static part; dynamic capture is performed, in which each dynamic part in the frame image after the static optimization is completed is analyzed, the corresponding dynamic part of each frame image is collected, and each frame image is played in a dynamic form, and the corresponding dynamic part of each frame image is analyzed to obtain the dynamic action in the real-time video image, and the dynamic action in the real-time video image is analyzed to determine whether the current dynamic action is normal.

[0058] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A dynamic capture analysis algorithm based on video images, characterized in that: The specific steps of dynamic capture analysis are as follows: S1: Image analysis: The video is acquired through the real-time monitoring device and marked as a real-time video image. After the real-time video image is acquired, the corresponding real-time video image is analyzed to determine whether the quality of the real-time video image is qualified; S2: Region segmentation: perform region segmentation on the video image that has passed the image analysis, divide the video image into i frames, where i is a natural number greater than 1, and collect pixels in each frame. The distribution of the pixels in the frame is obtained through the collected pixels, and feature comparison is performed on the region of the frame where the pixel is located. Region segmentation is performed based on the feature comparison, and each frame is divided into o sub-regions. S3: Frame reduction optimization, analyzing the frame pictures after the video image is divided, determining whether the picture of the current frame is static, optimizing the static part, and reducing the resolution of the static part; the specific steps of frame reduction optimization in step S3 are as follows: Step S31: Arrange the frame pictures corresponding to the video image in the order in which the video image is divided, analyze the analysis sub-area in each frame picture, obtain the number of subjects in the analysis sub-area and the pixel position of the corresponding subject, and analyze the number of subjects in the corresponding analysis sub-area in adjacent frame pictures and the pixel position of the corresponding subject; Step S32: If there is a numerical fluctuation in the number of subjects in the corresponding analysis sub-region in the adjacent frame pictures, or there is a pixel position change in the pixel position of the corresponding subject in the corresponding analysis sub-region in the adjacent frame pictures, it is determined that there is a dynamic operation in the current analysis sub-region, and the corresponding analysis sub-region is marked as a dynamic sub-region; if there is no numerical fluctuation in the number of subjects in the corresponding analysis sub-region in the adjacent frame pictures, and there is no pixel position change in the pixel position of the corresponding subject in the corresponding analysis sub-region in the adjacent frame pictures, it is determined that there is no dynamic operation in the current analysis sub-region, and the corresponding analysis sub-region is marked as a static sub-region; Step S33: controlling the resolution of the static sub-region in the corresponding frame image, and the resolution control standard is that the resolution of the static sub-region meets the normal image resolution threshold, and the resolution of the static sub-region is smaller than the resolution of the dynamic sub-region; S4: Dynamic capture, analyzing each dynamic part in the frame image that has completed static optimization, collecting the corresponding dynamic part of each frame image, playing each frame image in a dynamic form, analyzing the corresponding dynamic part of each frame image, acquiring the dynamic action in the real-time video image, and analyzing the dynamic action in the real-time video image to determine whether the current dynamic action is normal.

2. The video image-based dynamic capture analysis algorithm according to claim 1, characterized in that: The specific steps of image analysis in step S1 are as follows: S11: The video is acquired through a real-time monitoring device, marked as a real-time video image, and the quality of the video image is analyzed; S12: The average number of frames per second of the real-time video image and the resolution of the corresponding frame image are collected, and the average number of frames per second of the real-time video image and the resolution of the corresponding frame image are marked as ZS and FB respectively; through the formula The quality analysis coefficient X of the real-time video image is obtained, wherein a1 and a2 are both preset proportional coefficients, and a1 is greater than a2 and greater than 0; S13: Compare the quality analysis coefficient X of the real-time video image with the quality analysis coefficient threshold: If the quality analysis coefficient X of the real-time video image exceeds the quality analysis coefficient threshold, the corresponding real-time video image quality analysis is judged to be qualified, a quality analysis qualified signal is generated and the video image is marked as a qualified video image, and the process proceeds to step S2; if the quality analysis coefficient X of the real-time video image does not exceed the quality analysis coefficient threshold, the corresponding real-time video image quality analysis is judged to be unqualified, a quality analysis unqualified signal is generated and the video image is marked as an unqualified video image, and no motion capture is performed.

3. The video image-based dynamic capture analysis algorithm according to claim 1, characterized in that: The specific steps of region segmentation in step S2 are as follows: Step S21: segmenting the qualified video image into regions, analyzing the frame images divided into the video image, and performing feature comparison based on the analysis of the pixels in each frame image; Step S22: setting a pixel number threshold, and dividing the region in the frame image formed by the pixel number threshold into o sub-regions, and the sub-regions corresponding to the pixel number threshold have the same features, and the sub-region features are represented by the brightness and saturation of the corresponding sub-regions in the frame image; Step S23: If the corresponding values ​​of the data in the features in adjacent sub-regions are all within the corresponding data value threshold range, then the adjacent sub-regions are determined to be marked as sub-regions with the same features; if the corresponding values ​​of the data in the features in the adjacent sub-regions are not all within the corresponding data value threshold range, then the adjacent sub-regions are determined to be marked as sub-regions with different features; the corresponding sub-regions with the same features in the frame image are regularized, and the regularized sub-regions and sub-regions with different features are uniformly marked as analysis sub-regions, and the features in the same analysis sub-region are consistent, and the features in adjacent analysis sub-regions are inconsistent.

4. The video image-based dynamic capture analysis algorithm according to claim 1, characterized in that: The specific steps of dynamic capture in step S4 are as follows: Step S41: obtaining a frame image with a dynamic sub-region, analyzing the corresponding frame image, playing the frame image with the dynamic sub-region in a dynamic form, obtaining a position change trajectory of the corresponding subject in the dynamic sub-region through the change of the position of the corresponding subject pixel point in the dynamic sub-region, and marking the corresponding position change trajectory as a real-time subject action; Step S42: marking the corresponding dynamic sub-region as an action analysis region, limiting the action of the corresponding subject in the action analysis region, and marking the action that completes the limitation as a preset subject action; Step S43: comparing the real-time subject motion of the corresponding subject in the motion analysis area with the preset subject motion. If the real-time subject motion is consistent with the preset subject motion, the real-time motion of the corresponding subject is determined to be abnormal, an abnormal motion signal is generated and sent to the mobile phone terminal of the motion capture executor; If the real-time subject action is not consistent with the preset subject action, the real-time action of the corresponding subject is determined to be normal, a normal action signal is generated and sent to the mobile phone terminal of the motion capture executor.

Images