Personnel behavior monitoring method and device based on machine vision

Abstract

The invention provides a machine vision-based personnel behavior monitoring method and device. The machine vision-based personnel behavior monitoring method comprises the following steps of: acquiring operation human body data of a monitoring area; preprocessing the operation human body data to obtain a preprocessing result; the preprocessing at least comprises human body segmentation; extracting a skeleton line by using an angular bisector feature extraction algorithm according to the preprocessing result, and calculating angle and direction information of human joints to obtain a feature extraction result; and performing behavior analysis and recognition according to the feature extraction result. By using the angular bisector feature extraction method, the outermost layer of the graph is stripped until only one connected unit width skeleton is left, so that a more accurate and continuous skeleton line is obtained, and the accuracy of follow-up behavior monitoring is improved. Different behaviors can be accurately identified and classified by analyzing information such as movement, postures and behavior modes of the human skeleton line. Only the acquired operation human body data is used for analysis, a sensor does not need to be configured, the scene applicability is high, and comfort and convenience are good.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a method and device for monitoring human behavior based on machine vision. Background Art

[0002] In some scenarios, some behaviors of user objects may have an important impact on safe production, safe operation or safe operation within the scenario. Monitoring personnel behavior plays an important role in ensuring safety, improving efficiency and maintaining order.

[0003] Currently, specific sensors are usually used to monitor human behavior, such as inertial sensors, sound sensors, heart rate sensors, etc. These sensors may require additional equipment or wearables, and may affect the comfort and convenience of users. In addition, data fusion and consistency between different sensors may also affect accuracy. At the same time, using sensors to monitor human behavior requires user cooperation and participation, such as wearing sensors, performing specific actions, or answering questions. This may affect the user's behavior and limit the authenticity and accuracy of the monitoring results. There may also be differences in the degree of cooperation and willingness of users to participate.

[0004] In addition, sensor-based human behavior monitoring has limited application scenarios. For example, methods based on sound sensors may be disturbed in noisy environments, resulting in reduced recognition accuracy.

[0005] In summary, the existing technology has the problems of low scenario applicability, poor comfort and convenience, and low accuracy. Summary of the invention

[0006] The present invention provides a personnel behavior monitoring method and device based on machine vision, which are used to solve the defects of low scene applicability, poor comfort and convenience, and low accuracy in the prior art, and to achieve personnel behavior monitoring with high scene applicability, good comfort and convenience, and high accuracy.

[0007] The present invention provides a method for monitoring human behavior based on machine vision, comprising the following steps: Obtaining human body data of workers in the monitoring area; Preprocessing the operating human body data to obtain a preprocessing result; the preprocessing at least includes human body segmentation; According to the preprocessing result, the skeleton line is extracted by using the angle bisector feature extraction algorithm and the angle and direction information of the human body joints are calculated to obtain the feature extraction result; Behavior analysis and recognition are performed based on the feature extraction results.

[0008] According to a machine vision-based personnel behavior monitoring method provided by the present invention, the operating human body data is preprocessed to obtain a preprocessing result, which specifically includes: Performing image denoising on the operating human body data to obtain a denoising result; The human body is segmented according to the denoising result to obtain a preprocessing result.

[0009] According to a human behavior monitoring method based on machine vision provided by the present invention, skeleton lines are extracted using an angle bisector feature extraction algorithm according to the preprocessing results and the angle and direction information of human joints are calculated to obtain feature extraction results, which specifically includes: Performing corner point detection according to the preprocessing result; Determine an angle bisector according to the edge directions on both sides of the corner point, and determine a skeleton line according to the angle bisector; The angle and direction information of the human body joints are calculated according to the skeleton line to obtain a feature extraction result.

[0010] According to a method for monitoring personnel behavior based on machine vision provided by the present invention, behavior analysis and identification are performed according to the feature extraction result, specifically comprising: Analyze the motion, posture and behavior pattern of the human skeleton line according to the feature extraction result to obtain the analysis result; Different human behaviors are identified or classified according to the analysis results to obtain human behavior recognition results.

[0011] According to a human behavior monitoring method based on machine vision provided by the present invention, an angle bisector is determined according to the edge directions on both sides of the corner point, a skeleton line is determined according to the angle bisector, and then the method further includes: The feature extraction result is refined to remove noise points on the graphic boundary until the skeleton line reaches a corner with a specified minimum sharpness.

[0012] According to a machine vision-based personnel behavior monitoring method provided by the present invention, obtaining human body data of workers in a monitoring area specifically includes: The pre-set multiple image acquisition devices are used to dynamically or fixedly operate in the monitoring area to obtain the human body data to be operated.

[0013] The present invention also provides a human behavior monitoring device based on machine vision, comprising the following modules: A data unit is used to obtain the human body data of the workers in the monitoring area; A preprocessing unit, used for preprocessing the operation human body data to obtain a preprocessing result; the preprocessing at least includes human body segmentation; A feature unit, used to extract skeleton lines and calculate angle and direction information of human joints using an angle bisector feature extraction algorithm according to the preprocessing result to obtain a feature extraction result; A processing unit is used to perform behavior analysis and recognition based on the feature extraction result.

[0014] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for monitoring personnel behavior based on machine vision as described above is implemented.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the human behavior monitoring method based on machine vision as described in any one of the above is implemented.

[0016] The present invention also provides a computer program product, including a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned methods for monitoring human behavior based on machine vision.

[0017] The personnel behavior monitoring method and device based on machine vision provided by the present invention obtain the human body data of the workers in the monitoring area; pre-process the human body data to obtain the pre-processing results; the pre-processing at least includes human body segmentation; according to the pre-processing results, the skeleton line is extracted by using the angle bisector feature extraction algorithm and the angle and direction information of the human joints are calculated to obtain the feature extraction results; and behavior analysis and recognition are performed according to the feature extraction results. The present invention uses the angle bisector feature extraction method to peel off the outermost layer of the graphics until only a connected unit width skeleton is left, thereby obtaining a more accurate and continuous skeleton line, thereby improving the accuracy of subsequent behavior monitoring. By analyzing the movement, posture and behavior pattern of the human skeleton line and other information, different behaviors can be accurately identified and classified. The present invention only uses the acquired human body data for analysis, does not need to configure sensors, and has high scene applicability, comfort and convenience. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 It is a flow chart of the personnel behavior monitoring method based on machine vision provided by the present invention.

[0021] Figure 2 It is a structural schematic diagram of a personnel behavior monitoring device based on machine vision provided by the present invention.

[0022] Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not 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.

[0024] Combine the following Figure 1 Describe the personnel behavior monitoring method based on machine vision of the present invention, Figure 1 FIG. 1 is a flow chart of a method for monitoring personnel behavior based on machine vision provided by the present invention, such as Figure 1 As shown, the method comprises the following steps: Step 110: Acquire the human body data of the workers in the monitoring area.

[0025] It should be noted that the monitoring area refers to the area where the target object that needs to be monitored for behavior is located, and it can also be the area covered by the equipment that collects the operating human body data of the target object that needs to be monitored for behavior, or it can be a pre-delineated area. The present invention does not impose any restrictions on this.

[0026] In some embodiments, the monitoring area particularly includes densely populated areas such as offshore drilling platforms, campuses, cafeterias, and tourist attractions.

[0027] It is understandable that a large number of surveillance cameras are often installed in densely populated areas such as offshore drilling platforms, campuses, canteens, and tourist attractions. How to use computers to automatically identify dangerous actions and unsafe behaviors of a small number of people in densely populated areas is a necessary means to prevent accidents. Traditionally, methods such as monitoring personnel, increasing security personnel, and limiting flow are used to control, which is inefficient and difficult to capture accurately. The present invention describes how to use machine vision technology to achieve single-pixel level personnel behavior monitoring means, especially unsafe behaviors of personnel.

[0028] The working human body data includes image data and video data.

[0029] It should be noted that, in step 110, an image acquisition device is used to acquire the data of the human body working in the monitoring area. The present invention does not limit the specific image acquisition device, and the image acquisition device can be selected according to the actual situation.

[0030] In some embodiments, obtaining the human body data of the workers in the monitoring area specifically includes: The pre-set multiple image acquisition devices are used to dynamically or fixedly operate in the monitoring area to obtain the human body data to be operated.

[0031] Specifically, multiple fixed-position image acquisition devices may be arranged in the monitoring area to collect the data of the working body, or multiple pre-set image acquisition devices may be used to dynamically operate in the monitoring area to obtain the data of the working body.

[0032] Step 120: preprocessing the operating human body data to obtain a preprocessing result; the preprocessing at least includes human body segmentation.

[0033] In step 120, after obtaining the working human body data, the working human body data is preprocessed, and the preprocessing at least includes human body segmentation.

[0034] It is understood that human body segmentation refers to the process of separating the human body part in the image from the background. The present invention does not limit the specific algorithm of human body segmentation, and may include the following methods: 1) Threshold-based segmentation: distinguish foreground and background by setting thresholds such as brightness and color.

[0035] 2) Edge-based segmentation: Detect the edges of human bodies in the image and then perform segmentation through edge connections.

[0036] 3) Region-based segmentation: Segment the human body through techniques such as region growing and region merging.

[0037] 4) Graph-based segmentation: Using graph theory methods, the image is regarded as a graph and segmented using minimum cut algorithms, etc.

[0038] 5) Deep learning-based methods: such as using fully convolutional neural networks (FCN), Mask R-CNN, U-Net and other models for pixel-level segmentation.

[0039] Furthermore, the preprocessing also includes image denoising. In some embodiments, the operating human body data is preprocessed to obtain a preprocessing result, which specifically includes: Performing image denoising on the operating human body data to obtain a denoising result; The human body is segmented according to the denoising result to obtain a preprocessing result.

[0040] Specifically, the present invention does not limit the specific algorithm for performing image denoising. For example, image denoising can be achieved by using threshold processing.

[0041] Furthermore, the preprocessing also includes enhancement, image correction, etc. In one embodiment, the image data is subjected to image denoising, enhancement, image correction processing and human body segmentation in sequence.

[0042] It should be noted that the present invention does not limit the specific algorithm for performing the above preprocessing. For example, grayscale transformation can be used to achieve image enhancement, and radiometric correction can be used to achieve image correction.

[0043] Step 130: According to the preprocessing result, the skeleton line is extracted using the angle bisector feature extraction algorithm and the angle and direction information of the human body joints are calculated to obtain the feature extraction result.

[0044] In step 130, the angle bisector and skeleton line features of the human body image are extracted to calculate the angle and direction information of the human body joints. It is understandable that the traditional skeleton extraction algorithm usually needs to start from the outermost layer of the figure and gradually peel off until only a connected unit-width skeleton remains. This often has some limitations and it is difficult to accurately capture the skeletal structure of the target object. The present invention solves the problem of peeling off the outermost layer of the figure until only a connected unit-width skeleton remains by using the angle bisector feature extraction method. In this way, a more accurate and continuous skeleton line can be obtained, thereby improving the accuracy of subsequent behavior monitoring. At the same time, the angle bisector feature can more accurately capture the position and direction information of the target joint point, providing more reliable basic data for subsequent posture estimation and behavior analysis.

[0045] It should be explained that the feature extraction result includes the extracted skeleton line and the angle and direction information of the human body joints.

[0046] Step 140: Perform behavior analysis and recognition based on the feature extraction results.

[0047] In step 140, the extracted skeleton line is used in combination with the angle and direction information of the human joints to perform specific behavior analysis and recognition. It can be understood that the present invention adds a diagnostic function on the basis of the traditional machine vision detection method. Therefore, by analyzing the movement, posture and behavior pattern of the human skeleton line, different behaviors can be accurately identified and classified, and there is no need to configure sensors, which has high scene applicability, comfort and convenience.

[0048] It should be noted that in dangerous environments such as fire and explosion, the evacuation behavior of the crowd is difficult to capture and monitor due to emergency response, especially the dangerous evacuation method is difficult to control, which may lead to crowding, trampling, and even secondary injury incidents. The personnel behavior monitoring method based on machine vision provided by the present invention can monitor the emergency evacuation status of personnel in real time without increasing any cost (such as traditional wearable devices). The present invention can provide technical support for on-site emergency command and safety assurance.

[0049] The following further explains step 130. In some embodiments, according to the preprocessing result, the skeleton line is extracted using the angle bisector feature extraction algorithm and the angle and direction information of the human body joints are calculated to obtain the feature extraction result, which specifically includes: Performing corner point detection according to the preprocessing result; Determine an angle bisector according to the edge directions on both sides of the corner point, and determine a skeleton line according to the angle bisector; The angle and direction information of the human body joints are calculated according to the skeleton line to obtain a feature extraction result.

[0050] Specifically, corner detection is first performed based on the preprocessing result. The specific steps include: calculating the gradient of each pixel in the preprocessing result image to determine the intensity change of each point in the image; using an algorithm (such as Harris corner detection or Shi-Tomasi algorithm) to find the corners in the image.

[0051] After obtaining the corner points, for each detected corner, the edge directions of both sides are calculated, and then the angle bisector of the two sides is determined. After that, the skeleton line is determined based on the angle bisector. Specifically, in a specific embodiment, the steps of determining the skeleton line include: obtaining the color and depth of the monitored object in the scene; extracting 100 frames of data of various parts of the human body; extracting the features of the skeleton information in the 100 frames of data; obtaining the color information of the joints of the human body; numbering the various colors in the scene; marking the colors of the skeleton parts in the 100 frames of images, determining the angle bisector based on the marking results, and then determining the skeleton line.

[0052] Furthermore, an implementation example is given for determining the angle bisector based on the annotation results and then determining the skeleton line: for each skeletal joint, find the two joint points adjacent to it; calculate the angle formed by these two joint points and the central joint point; find the angle bisector of this angle, and according to the direction vector of the angle bisector, start from the central joint point and extend along the direction of the angle bisector until encountering another joint point, and connect these line segments to form a skeleton line.

[0053] Based on the obtained skeleton line, the angle and direction information of the human body joints are calculated to obtain the feature extraction results.

[0054] Furthermore, in some embodiments, an angle bisector is determined according to edge directions on both sides of the corner point, and a skeleton line is determined according to the angle bisector, and then the method further includes: The feature extraction result is refined to remove noise points on the graphic boundary until the skeleton line reaches a corner with a specified minimum sharpness.

[0055] Specifically, the extracted skeleton line is thinned to remove noise points on the boundary of the figure and retain the main skeleton line axis. The specific steps include: first identifying the boundary points of the figure and detecting the corner points of the figure to ensure that the shape of the figure will not be changed during the thinning process, and then iteratively removing the points on the boundary until the skeleton line reaches the corner with the specified minimum sharpness, and the remaining figure is a skeleton line of unit width.

[0056] The embodiment of the present invention adopts a more optimized thinning algorithm, which can effectively remove points on the boundary of the graphic, that is, effectively peel off the outermost layer of the graphic in image processing until a connected unit width skeleton line is formed, thereby improving the processing speed of subsequent steps and reducing the amount of data while retaining the shape information of the image, which has important application value.

[0057] The following further explains step 140. In some embodiments, the behavior analysis and recognition are performed according to the feature extraction result, specifically including: Analyze the motion, posture and behavior pattern of the human skeleton line according to the feature extraction result to obtain the analysis result; Different human behaviors are identified or classified according to the analysis results to obtain human behavior recognition results.

[0058] Specifically, based on the extracted and refined skeleton line, specific behavior analysis and recognition are performed. Specifically, during the implementation process, by analyzing the movement, posture and behavior pattern of the human skeleton line, different behaviors can be accurately identified and classified to obtain the analysis results of personnel behavior monitoring. Then, different human behaviors are identified or classified according to the analysis results.

[0059] It should be noted that the embodiment of the present invention does not limit the specific algorithm for analyzing the motion, posture and behavior pattern of the human skeleton line. In some embodiments, the analysis of the motion, posture and behavior pattern of the human skeleton line includes the following steps: The feature extraction results within a preset time range are used to form a feature extraction result sequence. Based on the feature extraction result sequence, a time series analysis algorithm is used to identify the behavior sequence. The time series analysis algorithm includes a hidden Markov model (HMM) or a recurrent neural network (RNN). Furthermore, a machine learning algorithm (such as a clustering or classification algorithm) is used to identify and classify the behavior pattern based on the behavior sequence.

[0060] Furthermore, the present invention does not limit the specific algorithm for identifying or classifying different human behaviors based on the analysis results. In some embodiments, identifying or classifying different human behaviors based on the analysis results includes the following steps: constructing a feature vector for behavior recognition from the extracted skeleton lines, joint angles and direction information, constructing a data set and marking it; selecting a machine learning model suitable for behavior recognition for model training, and the trained model can be used to identify or classify human behaviors.

[0061] Among them, machine learning models include but are not limited to: support vector machine (SVM), random forest (RandomForest), gradient boosting decision trees (Gradient Boosting Trees), neural networks (Neural Networks), convolutional neural networks (CNN), and recurrent neural networks (RNN).

[0062] The present invention extracts and analyzes the angle bisector feature, and can monitor the behavior of personnel.

[0063] The personnel behavior monitoring method based on machine vision provided by the present invention obtains the human body data of the workers in the monitoring area; pre-processes the human body data to obtain the pre-processing results; the pre-processing at least includes human body segmentation; according to the pre-processing results, the skeleton line is extracted by using the angle bisector feature extraction algorithm and the angle and direction information of the human joints are calculated to obtain the feature extraction results; and behavior analysis and recognition are performed according to the feature extraction results. The present invention uses the angle bisector feature extraction method to peel off the outermost layer of the graphics until only a connected unit width skeleton remains, thereby obtaining a more accurate and continuous skeleton line, thereby improving the accuracy of subsequent behavior monitoring. By analyzing the information such as the movement, posture and behavior pattern of the human skeleton line, different behaviors can be accurately identified and classified. The present invention only uses the acquired human body data for analysis, does not need to configure sensors, and has high scene applicability, good comfort and convenience.

[0064] The following is a description of the personnel behavior monitoring device based on machine vision provided by the present invention. The personnel behavior monitoring device based on machine vision described below and the personnel behavior monitoring method based on machine vision described above can be referred to each other. Figure 2 As shown, the device includes the following modules: The data unit 210 is used to obtain the human body data of the workers in the monitoring area; A preprocessing unit 220 is used to preprocess the operating human body data to obtain a preprocessing result; the preprocessing at least includes human body segmentation; A feature unit 230 is used to extract skeleton lines and calculate angle and direction information of human joints using an angle bisector feature extraction algorithm according to the preprocessing result to obtain a feature extraction result; The processing unit 240 is used to perform behavior analysis and recognition according to the feature extraction result.

[0065] According to a machine vision-based personnel behavior monitoring device provided by the present invention, the operating human body data is preprocessed to obtain a preprocessing result, which specifically includes: Performing image denoising on the operating human body data to obtain a denoising result; The human body is segmented according to the denoising result to obtain a preprocessing result.

[0066] According to a machine vision-based personnel behavior monitoring device provided by the present invention, skeleton lines are extracted using an angle bisector feature extraction algorithm according to the preprocessing result, and angle and direction information of human joints are calculated to obtain feature extraction results, which specifically includes: Performing corner point detection according to the preprocessing result; Determine an angle bisector according to the edge directions on both sides of the corner point, and determine a skeleton line according to the angle bisector; The angle and direction information of the human body joints are calculated according to the skeleton line to obtain a feature extraction result.

[0067] According to a machine vision-based personnel behavior monitoring device provided by the present invention, behavior analysis and recognition are performed according to the feature extraction result, specifically including: Analyze the motion, posture and behavior pattern of the human skeleton line according to the feature extraction result to obtain the analysis result; Different human behaviors are identified or classified according to the analysis results to obtain human behavior recognition results.

[0068] According to a machine vision-based personnel behavior monitoring device provided by the present invention, an angle bisector is determined according to the edge directions on both sides of the corner point, and a skeleton line is determined according to the angle bisector, and then the following further comprises: The feature extraction result is refined to remove noise points on the graphic boundary until the skeleton line reaches a corner with a specified minimum sharpness.

[0069] According to a machine vision-based personnel behavior monitoring device provided by the present invention, the data of human bodies working in the monitoring area is obtained, specifically including: The pre-set multiple image acquisition devices are used to dynamically or fixedly operate in the monitoring area to obtain the human body data to be operated.

[0070] The personnel behavior monitoring device based on machine vision provided by the present invention obtains the human body data of the workers in the monitoring area; pre-processes the human body data to obtain a pre-processing result; the pre-processing at least includes human body segmentation; according to the pre-processing result, the skeleton line is extracted using the angle bisector feature extraction algorithm and the angle and direction information of the human joints are calculated to obtain a feature extraction result; and behavior analysis and recognition are performed according to the feature extraction result. The present invention uses the angle bisector feature extraction method to peel off the outermost layer of the graphic until only a connected unit width skeleton remains, thereby obtaining a more accurate and continuous skeleton line, thereby improving the accuracy of subsequent behavior monitoring. By analyzing the information such as the movement, posture and behavior pattern of the human skeleton line, different behaviors can be accurately identified and classified. The present invention only uses the acquired human body data for analysis, does not need to configure sensors, and has high scene applicability, good comfort and convenience.

[0071] Figure 3 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 3 As shown, the electronic device may include: a processor 310, a communication interface 320, a memory 330 and a communication bus 340, wherein the processor 310, the communication interface 320 and the memory 330 communicate with each other through the communication bus 340. The processor 310 may call the logic instructions in the memory 330 to execute a personnel behavior monitoring method based on machine vision, the method comprising: obtaining operating human body data in a monitoring area; preprocessing the operating human body data to obtain a preprocessing result; the preprocessing at least includes human body segmentation; extracting skeleton lines and calculating the angle and direction information of human joints using an angle bisector feature extraction algorithm according to the preprocessing result to obtain a feature extraction result; and performing behavior analysis and recognition according to the feature extraction result.

[0072] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0073] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the personnel behavior monitoring method based on machine vision provided by the above-mentioned methods, and the method includes: acquiring operating human body data in a monitored area; preprocessing the operating human body data to obtain a preprocessing result; the preprocessing at least includes human body segmentation; according to the preprocessing result, using an angle bisector feature extraction algorithm to extract skeleton lines and calculate the angle and direction information of human joints to obtain a feature extraction result; and performing behavior analysis and recognition according to the feature extraction result.

[0074] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the machine vision-based personnel behavior monitoring method provided by the above-mentioned methods, the method comprising: acquiring human body data of workers in a monitoring area; preprocessing the human body data to obtain a preprocessing result; the preprocessing at least includes human body segmentation; based on the preprocessing result, using an angle bisector feature extraction algorithm to extract skeleton lines and calculate the angle and direction information of human joints to obtain a feature extraction result; and performing behavior analysis and recognition based on the feature extraction result.

[0075] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0076] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for monitoring personnel behavior based on machine vision, characterized in that: include: Obtaining human body data of workers in the monitoring area; Preprocessing the operating human body data to obtain a preprocessing result; The preprocessing includes at least human body segmentation; According to the preprocessing result, the skeleton line is extracted by using the angle bisector feature extraction algorithm and the angle and direction information of the human body joints are calculated to obtain the feature extraction result; Behavior analysis and recognition are performed based on the feature extraction results.

2. The method for monitoring human behavior based on machine vision according to claim 1, characterized in that: Preprocessing the operating human body data to obtain preprocessing results, specifically including: Performing image denoising on the operating human body data to obtain a denoising result; The human body is segmented according to the denoising result to obtain a preprocessing result.

3. The method for monitoring human behavior based on machine vision according to claim 1, characterized in that: According to the preprocessing results, the skeleton line is extracted using the angle bisector feature extraction algorithm and the angle and direction information of the human joints are calculated to obtain the feature extraction results, which specifically include: Performing corner point detection according to the preprocessing result; Determine an angle bisector according to the edge directions on both sides of the corner point, and determine a skeleton line according to the angle bisector; The angle and direction information of the human body joints are calculated according to the skeleton line to obtain a feature extraction result.

4. The method for monitoring human behavior based on machine vision according to claim 1, characterized in that: Behavior analysis and identification are performed based on the feature extraction results, specifically including: Analyze the motion, posture and behavior pattern of the human skeleton line according to the feature extraction result to obtain the analysis result; Different human behaviors are identified or classified according to the analysis results to obtain human behavior recognition results.

5. The method for monitoring human behavior based on machine vision according to claim 1, characterized in that: An angle bisector is determined according to the edge directions on both sides of the corner point, and a skeleton line is determined according to the angle bisector, and then the method further includes: The feature extraction result is refined to remove noise points on the graphic boundary until the skeleton line reaches a corner with a specified minimum sharpness.

6. The method for monitoring human behavior based on machine vision according to claim 1, characterized in that: Obtain human body data in the monitoring area, including: The pre-set multiple image acquisition devices are used to dynamically or fixedly operate in the monitoring area to obtain the human body data to be operated.

7. A personnel behavior monitoring device based on machine vision, characterized in that: include: A data unit is used to obtain the human body data of the workers in the monitoring area; A preprocessing unit, used for preprocessing the operating human body data to obtain a preprocessing result; The preprocessing includes at least human body segmentation; A feature unit, used to extract skeleton lines and calculate angle and direction information of human joints using an angle bisector feature extraction algorithm according to the preprocessing result to obtain a feature extraction result; A processing unit is used to perform behavior analysis and recognition based on the feature extraction result.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for monitoring personnel behavior based on machine vision as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for monitoring human behavior based on machine vision as claimed in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for monitoring human behavior based on machine vision as claimed in any one of claims 1 to 6 is implemented.

Images