Station area video monitoring patrol method and system, terminal and storage medium

Abstract

The invention relates to the technical field of station area monitoring, in particular to a station area video monitoring patrol method and system, a terminal and a storage medium, and the method comprises the steps: presetting a patrol area and a key area, and collecting target basic data and target image data in the patrol area in real time; judging whether the target falls into the key area based on a radar detection algorithm in combination with the target basic data, and marking the target as a suspected target when judging that the target falls into the key area; when the target falls into the key area, target tracking is carried out, and real-time target trajectory information is acquired; target detection is carried out, and clothes feature information of a target is obtained; and matching the clothing feature information with a preset sample library, judging whether the target has a wandering behavior based on the target trajectory information, triggering an alarm mechanism when the clothing feature information of the target is not matched and / or the wandering behavior exists, and recording alarm information. The intelligent and precise level of station area safety management is effectively improved, the accident risk is reduced, and the overall operation safety is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of station area monitoring, and in particular to a station area video monitoring patrol method, system, terminal and storage medium. Background Art

[0002] The station area usually refers to the area including a specific station (railway) and surrounding related areas. Station area isolation is a measure to separate the station area from the external environment by physical or technical means, or to separate different functional areas and areas with different security levels within the station area.

[0003] The station area isolation is mainly protected by physical means, generally using fences, barbed cages, etc., and is equipped with warning signs to ensure the safe and orderly operation of the station area, control the flow of personnel, and prevent illegal intrusion.

[0004] When physical defense measures are adopted, the fence may be damaged due to wind, sun, human vandalism, etc., and needs to be inspected and repaired regularly. In addition, when the fence is damaged and has holes, it is easy for people to enter the station area, causing driving and personal safety problems. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method, a system, a terminal and a storage medium to solve the above-mentioned technical problems.

[0006] In a first aspect, the present invention provides a station area video surveillance patrol method, comprising the following steps: S1, based on the patrol needs of the station area, preset patrol areas and key areas within the patrol areas, and collect basic data and target image data within the patrol areas in real time; S2, based on the radar detection algorithm combined with the target basic data, determines whether the target falls into the key area. If the target is determined to fall into the key area, the target is marked as a suspected target; S3, when the target falls into the key area, the target is tracked based on the data fusion algorithm combined with the target basic data and target image data to obtain real-time target trajectory information; Perform target detection on target image data based on convolutional neural network algorithm and obtain target clothing feature information; S4, matching the clothing feature information with a preset sample library, judging whether the target has wandering behavior based on the target trajectory information, triggering an alarm mechanism when the target's clothing feature information does not match and / or has wandering behavior, and recording the alarm information.

[0007] In an optional embodiment, in step S3, the convolutional neural network algorithm specifically includes: a backbone network, a neck network and a head network; In the backbone network, the scattered feature information is integrated based on the long-range attention mechanism. At the same time, a 3×3 convolution kernel is used in the first convolution layer, and parallel plus serial max pooling is adopted. The neck network adopts a decoupled head design, which processes target recognition and target feature extraction through independent branches; The head network uses the Anchor Free mode to directly predict the candidate regions of the target. For each predicted target candidate region, its classification score and regression score are calculated, and the correct target candidate region is determined based on the classification score and regression score. The weighted comprehensive BCE Loss, Distribution Focal Loss and CIoU Loss are used as the loss function of the convolutional neural network algorithm. The loss value is transmitted back to the convolutional neural network algorithm through the back propagation algorithm and the parameters are updated.

[0008] In an optional embodiment, in step S3, obtaining the target's clothing feature information includes: extracting the target's upper body clothing area using an image segmentation algorithm, specifically including: Based on the top-down method, a convolutional neural network is used to obtain high-level semantic features from image data to preliminarily locate the upper body clothing area; the high-level semantic features are gradually mapped back to the original image resolution, and upsampling and convolution operations are used to accurately adjust the clothing area boundaries; The original pixel-level features of the image data are obtained based on the bottom-up method, and the local features and boundary information of the target are obtained by using clustering and edge detection algorithms to construct the upper body clothing area candidates; The upper body clothing area obtained by the Top-down method and the upper body clothing area obtained by the Bottom-up method are merged to obtain the final upper body clothing area.

[0009] In an optional embodiment, the process of fusing the upper body clothing region obtained by the top-down method and the upper body clothing region obtained by the bottom-up method specifically includes: For the areas that are consistent between the two results, they are directly determined as the reliable parts of the upper body clothing area; For areas with differences, check the correlation between the difference area and the surrounding reliable parts. If the difference area should belong to the same whole as an adjacent reliable part, merge the difference area. If the difference area does not belong to the same whole as an adjacent reliable part, mark it separately.

[0010] In an optional embodiment, the processing of the area with differences further includes using human body posture and prior knowledge to correct: Based on the human posture estimation results and the prior knowledge of human body structure, the relationship between the difference area and various parts of the human body is evaluated; If the difference area is located near the human body's motion joints and its segmentation difference affects the rationality of the human body posture, the area is adjusted according to the deformation law of clothing under the normal motion posture of the human body.

[0011] In an optional implementation, determining whether the target has wandering behavior based on the target trajectory information in step S4 specifically includes: Based on the target trajectory information, the target's trajectory length, number of movement direction changes, and residence time in a certain area within a preset time are calculated; If the length of the pedestrian's trajectory is lower than the preset length threshold, the direction of movement changes multiple times, and the time spent in the key area exceeds the preset actual threshold, the pedestrian is judged to be wandering. In an optional implementation, the alarm information in step S4 includes the number of alarms, the type of alarms and the locations where the alarms frequently occur. The number of alarms, the type of alarms and the locations where the alarms frequently occur are formed into a report and stored according to a preset upload interval.

[0012] In a second aspect, the present invention provides a station area video surveillance patrol system, which implements the above-mentioned station area video surveillance patrol method when executed, and the system includes: The data acquisition module presets patrol areas and key areas in patrol areas based on the patrol needs of the station area, and collects basic data and image data of targets in the patrol areas in real time; The radar judgment module determines whether the target falls into the key area based on the radar detection algorithm and the target basic data. If the target falls into the key area, the target is marked as a suspected target; Target tracking and feature extraction module: When a target falls into the key area, it tracks the target based on the data fusion algorithm and combines the target basic data and target image data to obtain real-time target trajectory information; Perform target detection on target image data based on convolutional neural network algorithm and obtain target clothing feature information; The alarm module matches the clothing feature information with the preset sample library, and determines whether the target has wandering behavior based on the target trajectory information. When the target's clothing feature information does not match and / or there is wandering behavior, the alarm mechanism is triggered and the alarm information is recorded.

[0013] In a third aspect, a terminal is provided, including: processor, memory, wherein: The memory is used to store computer programs. The processor is used to call and run the computer program from the memory, so that the terminal executes the above-mentioned terminal method.

[0014] According to a fourth aspect, a computer storage medium is provided, wherein the computer-readable storage medium stores instructions, and when the instructions are executed on a computer, the computer executes the methods described in the above aspects.

[0015] The beneficial effects of the present invention lie in that the station area video surveillance patrol method, system, terminal and storage medium provided by the present invention, by presetting areas and collecting multi-source data in real time, use advanced algorithms to achieve accurate target monitoring and behavior analysis, and effectively improve the safety and management efficiency of the station area, which is specifically manifested in: presetting patrols and key areas and collecting data to provide a basis for comprehensive monitoring, so that the monitoring range is clear and the data is comprehensive; the radar algorithm combines basic data to determine the landing point of the target area, and timely marks suspected targets to help quickly focus on potential risks; data fusion tracking obtains trajectories, combines convolutional neural networks to detect clothing features, and achieves dynamic grasp of targets and accurate feature identification, which is convenient for in-depth analysis of target attributes; clothing feature matching and trajectory analysis determine wandering behavior, trigger alarms and record information, which can accurately prevent violations and abnormal behaviors, ensure the order of the station area, and provide a basis for subsequent investigations, effectively improve the intelligence and precision of station area safety management, reduce accident risks, and enhance overall operational safety.

[0016] In addition, the invention has a reliable design principle, a simple structure and a very broad application prospect. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 It is a schematic flow chart of a station area video surveillance patrol method according to an embodiment of the present invention.

[0019] Figure 2 It is a schematic block diagram of a station area video surveillance and patrol system according to an embodiment of the present invention.

[0020] Figure 3 A schematic diagram of the structure of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only 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 should fall within the scope of protection of the present invention.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0023] The station area video surveillance and patrol method provided in the embodiment of the present invention is executed by a computer device, and accordingly, the station area video surveillance and patrol system runs in the computer device.

[0024] Figure 1 is a schematic flow chart of a method according to an embodiment of the present invention. Figure 1 The execution subject can be a station area video monitoring patrol system. According to different requirements, the order of the steps in the flow chart can be changed, and some can be omitted.

[0025] like Figure 1 As shown, the method includes: S1, based on the patrol needs of the station area, preset patrol areas and key areas within the patrol areas, and collect basic data and target image data within the patrol areas in real time; Clear regional division makes the monitoring scope clear and resource allocation more targeted, avoiding the waste of resources caused by blind monitoring. Real-time collection of target basic data (such as radar detection data) and target image data (such as high-definition camera images) can obtain multi-dimensional information of targets in the station area in all directions.

[0026] S2, based on the radar detection algorithm combined with the target basic data, determines whether the target falls into the key area. If the target is determined to fall into the key area, the target is marked as a suspected target; The advantages of radar's high precision and long-range detection are fully utilized. It can quickly and accurately identify targets entering key areas and mark them as suspected targets in a timely manner. This process greatly improves the system's response speed to potential security threats.

[0027] S3, when the target falls into the key area, the target is tracked based on the data fusion algorithm combined with the target basic data and target image data to obtain real-time target trajectory information; Perform target detection on target image data based on convolutional neural network algorithm and obtain target clothing feature information; The advantages of radar and camera are integrated to make target tracking more stable, continuous and accurate. No matter what complex environment or motion state the target is in, the system can "lock" the target tightly and grasp its dynamic trajectory in real time, providing a reliable basis for judging the target's behavior. At the same time, the convolutional neural network algorithm is used to detect the target's clothing characteristics, which can deeply mine the target's detailed information, help identify the target's identity (such as distinguishing between staff and outsiders) and judge its behavioral intentions (such as whether it meets the dress code in a specific scene).

[0028] S4, matching the clothing feature information with a preset sample library, judging whether the target has wandering behavior based on the target trajectory information, triggering an alarm mechanism when the target's clothing feature information does not match and / or has wandering behavior, and recording the alarm information.

[0029] By comparing clothing features with the preset sample library, unauthorized personnel or personnel with improper clothing can be found in time to enter key areas, effectively preventing external illegal intrusion and internal illegal operations. At the same time, based on the target trajectory information, the wandering behavior can be accurately judged, and suspicious personnel can be quickly identified, further enhancing the early warning capability of potential security threats. After the alarm mechanism is triggered, the alarm information is recorded in time, providing complete and detailed information for subsequent investigation and processing, which helps to quickly trace the incident, find the cause, and investigate the responsibility, ensuring that the station area safety management work forms a closed loop, effectively ensuring the safety and order of the station area operation, and making the station area safety management more scientific, efficient and reliable.

[0030] To facilitate understanding of the present invention, the station area video surveillance patrol method provided by the present invention is further described below based on the principle of the station area video surveillance patrol method of the present invention and in combination with the process of managing patient diagnosis and treatment information in the embodiment.

[0031] Optionally, as an embodiment of the present invention, front-end detection and alarm equipment are deployed in the throat and important channel areas of the station. It includes installing 2-3 laser detection radars, with dimensions of 270315240mm in length, width and height, using flanges and reinforcing ribs (made of Q235B steel plates), the whole is galvanized and the surface is sprayed with plastic for corrosion protection (gray), fixed by pre-buried ground cages and 8.8-grade hot-dip galvanized bolts, the cabinet rear panel door is locked for anti-theft, the detection radius is not less than 50 meters, the accuracy error is ±30mm, the detection angle is not less than 210 degrees, and the detection zone can be custom drawn (the number is not less than 16), and the laser radar communication outputs the analysis and judgment results through the RJ45 interface Ethernet; install 2-3 sets of monitoring data communication modules and 4-8 smart high-definition cameras (the outdoor camera has a maximum resolution of 4 million pixels, 1920×1080, can output Full HD1080p@30fps real-time images, focal length 4.8-115mm, 23 Optical zoom, can effectively monitor and alarm in general rain, snow and foggy weather conditions, the shell protection level is not less than IP55), 4 - 8 alarm sound columns, 4 - 8 sets of power supply equipment, 4 - 8 sets of customized mounting brackets. These devices are used as the front-end data acquisition and on-site alarm triggering equipment of the system. The working power supply is AC220V±20%, the rated power is 150W, and they can work reliably in an environment where the atmospheric pressure is not less than 70kPa (the altitude does not exceed 3000m), the relative humidity of the air is not more than 90% (25℃), the ambient air temperature is -40℃~+50℃ (outdoor) -10℃~+40℃ (indoor), the vibration frequency is 10Hz~55Hz and the acceleration amplitude is 10m / s².

[0032] Optionally, as an embodiment of the present invention, the station entrances and exits, platforms, track throat areas, freight yards, parking lots, and various equipment and facility concentration areas are divided into different patrol sub-areas. In these patrol areas, key areas are further determined, such as the edge of the platform (to prevent people from falling onto the track), track throat areas (to ensure the safety of train operation), and freight yard entrances and exits (to ensure the safety of cargo loading and unloading). These key areas are usually core areas where safety accidents are frequent, have a significant impact on operations, or involve the flow of important equipment and personnel.

[0033] Use tools such as geographic information systems (GIS) or station layout maps to digitally mark and set parameters for preset patrol areas and key areas, including information such as area boundaries, areas, coordinate ranges, etc., so that subsequent systems can accurately identify and locate the target's position within the area.

[0034] The LiDAR detector deployed in the patrol area works continuously at a specific scanning frequency (such as 20Hz), emitting laser beams and receiving reflected signals. The distance between the target object and the radar is calculated by measuring the round-trip time of the laser beam. At the same time, the basic data such as the position coordinates (x, y, z) and relative speed (v) of the target object in three-dimensional space are determined based on the scanning angle and position information of the radar.

[0035] Intelligent high-definition cameras installed at various key locations synchronously collect video image data at a certain frame rate (such as 30fps). The camera has high resolution (4 million pixels, 1920×1080) and good optical zoom capability (23x optical zoom), which can clearly capture target details at different distances and angles. The camera's field of view has been carefully adjusted to ensure coverage of the entire patrol area, while focusing on key areas to obtain high-quality target image information. During the acquisition process, the camera automatically adapts to changes in ambient light and has a day and night mode switching function to ensure that image data can be effectively collected under different lighting conditions (such as strong light during the day and low light at night). The image data is transmitted to the platform central control system in the form of a digital video stream, providing rich visual information for subsequent target detection and analysis.

[0036] Optionally, as an embodiment of the present invention, the target basic data collected by the laser radar is transmitted to the platform central control system in real time, and the radar detection algorithm in the system processes these data. First, the data is filtered and denoised to remove false signals and noise points caused by environmental interference (such as bad weather, reflections from surrounding objects, etc.) to ensure the accuracy and reliability of the data.

[0037] Next, the algorithm associates and integrates the continuous radar scanning data based on the target's position coordinates and movement speed to form the target's movement trajectory fragments. By analyzing the characteristics of these trajectory fragments, such as position change trends, speed stability, and movement direction consistency, the target's movement pattern and behavioral intention can be preliminarily determined.

[0038] The system compares the processed target location information with the preset critical area boundaries and parameters. When the target's location coordinates fall within the critical area, the system marks the target as a suspected target. At the same time, the system records the time, location, and initial motion state of the suspected target entering the critical area to provide basic data for further analysis and tracking. For example, if the location coordinates of a target object are determined to be within the critical area at the edge of the platform, the system immediately marks it as a suspected target and records the specific time point and initial speed direction of its entry into the area, so that its behavior can be closely monitored to determine whether there are safety risks.

[0039] Optionally, as an embodiment of the present invention, in step S3, the convolutional neural network algorithm specifically includes: a backbone network, a neck network and a head network; In the backbone network, the scattered feature information is integrated based on the long-range attention mechanism. At the same time, a 3×3 convolution kernel is used in the first convolution layer, and parallel plus serial max pooling is adopted. The neck network adopts a decoupled head design, which processes target recognition and target feature extraction through independent branches; The head network uses the Anchor Free mode to directly predict the candidate regions of the target. For each predicted target candidate region, its classification score and regression score are calculated, and the correct target candidate region is determined based on the classification score and regression score. The weighted comprehensive BCE Loss, Distribution Focal Loss and CIoU Loss are used as the loss function of the convolutional neural network algorithm. The loss value is transmitted back to the convolutional neural network algorithm through the back propagation algorithm and the parameters are updated.

[0040] For the detected pedestrian targets, the deep learning algorithm is further used to analyze their clothing features. The image area containing the pedestrian is cropped out and sent to a special clothing feature analysis module (based on the 18-layer residual network Resnet). This module extracts and classifies the color, texture, style and other features of the pedestrian's clothing in the image to determine whether the pedestrian is wearing clothing that complies with the regulations of the station area (such as protective clothing for railway staff, uniforms for specific positions, etc.). The 5 convolution groups in the network gradually abstract and extract features from the image. Each convolution group contains operations such as convolution, batch normalization and activation function, and performs downsampling once. The convolution kernel uses a 7x7 size, and the input image resolution is scaled to 224x224 for processing. Finally, the pedestrian's clothing feature information is output to provide auxiliary basis for subsequent safety management and abnormal behavior judgment.

[0041] Optionally, as an embodiment of the present invention, in step S3, obtaining the target's clothing feature information includes: extracting the target's upper body clothing area using an image segmentation algorithm, specifically including: Based on the top-down method, a convolutional neural network is used to obtain high-level semantic features from image data to preliminarily locate the upper body clothing area; the high-level semantic features are gradually mapped back to the original image resolution, and upsampling and convolution operations are used to accurately adjust the clothing area boundaries; The original pixel-level features of the image data are obtained based on the bottom-up method, and the local features and boundary information of the target are obtained by using clustering and edge detection algorithms to construct the upper body clothing area candidates; The upper body clothing area obtained by the Top-down method and the upper body clothing area obtained by the Bottom-up method are merged to obtain the final upper body clothing area.

[0042] Optionally, as an embodiment of the present invention, in the process of fusing the upper body clothing region obtained by the Top-down method and the upper body clothing region obtained by the Bottom-up method, the process specifically includes: For the areas that are consistent between the two results, they are directly determined as the reliable parts of the upper body clothing area; For areas with differences, check the correlation between the difference area and the surrounding reliable parts that have been determined. If the difference area should belong to the same whole as a reliable part, the difference area will be merged. If the difference area should not belong to the same whole as a reliable part, it will be marked separately. For example, if the difference area is obviously different from the main clothing in color or texture, and its position is consistent with the common decorative element position on clothing (such as special patterns on the neckline and cuffs), it will be marked as an independent sub-area to analyze its characteristics in more detail later.

[0043] Optionally, as an embodiment of the present invention, the processing of the area with differences further includes using human body posture and prior knowledge to correct: Based on the human posture estimation results and the prior knowledge of human body structure, the relationship between the difference area and various parts of the human body is evaluated; If the difference area is located near the human body's motion joints and its segmentation difference affects the rationality of the human body posture, the area should be adjusted according to the deformation law of clothing under normal human body motion posture. For example, if the bottom-up method over-segments the shoulder area, resulting in the connection between the arm and the body not conforming to the stretching and wrinkling of clothing under normal human body posture, the shoulder area should be reasonably merged or corrected according to the human body posture information provided by the top-down method to ensure that the segmentation result conforms to the principles of human kinematics. Optionally, as an embodiment of the present invention, judging whether the target has wandering behavior based on the target trajectory information in step S4 specifically includes: Based on the target trajectory information, the target's trajectory length, number of movement direction changes, and residence time in a certain area within a preset time are calculated; The system records and analyzes the target's coordinates at each time point based on the target's continuous position information in the past 60 seconds (preset time), and accurately measures the actual distance the target moves during this period through the distance formula between two points or a more complex trajectory fitting algorithm.

[0044] By analyzing the target's position change sequence within a preset time, the number of turning points in its movement direction is determined. For example, if the target moves first to the left, then to the right, and then to the left in the station area channel, the system will count the number of these direction changes.

[0045] The system records the time points when the target enters and leaves the key area and accurately calculates the length of time it stays in the area.

[0046] If the length of the pedestrian's trajectory is lower than the preset length threshold, the direction of movement changes multiple times, and the time spent in the key area exceeds the preset actual threshold, that is, wandering people usually have a small range of activities in a limited area (short trajectory length), unclear direction of action (frequent changes in direction), and stay in a specific area for a long time (long stay time), then the pedestrian is judged to be wandering.

[0047] Optionally, as an embodiment of the present invention, the alarm information in step S4 includes the number of alarms, the type of alarms and the locations where the alarms frequently occur. The number of alarms, the type of alarms and the locations where the alarms frequently occur are formed into a report and stored according to a preset upload interval.

[0048] Reports present data in a clear and intuitive way, usually in a table format, with column headings such as number of alarms, alarm type, and alarm frequent locations. Each row corresponds to the alarm data summary for a specific time period. For example, a daily report may show the number of occurrences of various types of alarm events and the corresponding frequent locations in different time periods of the day. The generated reports are stored in the system's database for subsequent query and analysis at any time.

[0049] Optionally, as an embodiment of the present invention, the use of the alarm mechanism specifically includes: In railway stations, if a pedestrian is detected near the tracks and is not wearing railway protective clothing or other prescribed clothing, the system will sound an alarm, prompting monitoring personnel to pay attention to the person's behavior and prevent possible safety accidents.

[0050] In railway stations, if a person wanders in a passage for a long time, there may be a tendency for someone to approach or there may be other safety hazards. The system will promptly send out an alarm signal to notify security personnel to check and take appropriate measures, such as asking their purpose, guiding them to leave, or conducting further investigations.

[0051] In some embodiments, the station area video surveillance patrol system may include multiple functional modules composed of computer program segments. The computer programs of each program segment in the station area video surveillance patrol system may be stored in the memory of a computer device and executed by at least one processor to perform (see Figure 1 Description) The function of video surveillance and patrol in the station area.

[0052] In this embodiment, the station area video surveillance patrol system can be divided into multiple functional modules according to the functions it performs, such as Figure 2 As shown. The functional modules of the system may include: a data acquisition module, a radar judgment module, a target tracking and feature extraction module and an alarm module. The module referred to in the present invention refers to a series of computer program segments that can be executed by at least one processor and can complete fixed functions, which are stored in a memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0053] The data acquisition module presets patrol areas and key areas in patrol areas based on the patrol needs of the station area, and collects basic data and image data of targets in the patrol areas in real time; The radar judgment module determines whether the target falls into the key area based on the radar detection algorithm and the target basic data. If the target falls into the key area, the target is marked as a suspected target; Target tracking and feature extraction module: When a target falls into the key area, it tracks the target based on the data fusion algorithm and combines the target basic data and target image data to obtain real-time target trajectory information; Perform target detection on target image data based on convolutional neural network algorithm and obtain target clothing feature information; The alarm module matches the clothing feature information with the preset sample library, and determines whether the target has wandering behavior based on the target trajectory information. When the target's clothing feature information does not match and / or there is wandering behavior, the alarm mechanism is triggered and the alarm information is recorded.

[0054] All modules of the station area monitoring system work together, with data collection in the preset area laying the foundation, radar judgment accurately locking onto suspected targets, tracking and feature extraction realizing dynamic monitoring and feature mastering, and the alarm module intelligently identifying abnormalities and triggering alarms, thus improving the overall safety, management efficiency and emergency response capabilities of the station area, and effectively ensuring safe and orderly operations.

[0055] Figure 3 A schematic diagram of the structure of a terminal provided in an embodiment of the present invention, which terminal can be used to execute the station area video monitoring and patrol method provided in an embodiment of the present invention.

[0056] The terminal may include: a processor, a memory and a communication unit. These components communicate via one or more buses. Those skilled in the art will appreciate that the server structure shown in the figure does not limit the present invention. It may be a bus structure or a star structure, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0057] The memory can be used to store the execution instructions of the processor, and the memory can be implemented by any type of volatile or non-volatile storage terminal or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. When the execution instructions in the memory are executed by the processor, the terminal is enabled to execute some or all of the steps in the following method embodiments.

[0058] The processor is the control center of the storage terminal, which uses various interfaces and lines to connect various parts of the entire electronic terminal, and executes various functions and / or processes data of the electronic terminal by running or executing software programs and / or modules stored in the memory, and calling data stored in the memory. The processor can be composed of an integrated circuit (IC), for example, it can be composed of a single packaged IC, or it can be composed of multiple packaged ICs with the same or different functions. For example, the processor can only include a central processing unit (CPU). In the embodiment of the present invention, the CPU can be a single computing core or multiple computing cores.

[0059] The communication unit is used to establish a communication channel so that the storage terminal can communicate with other terminals, receive user data sent by other terminals or send user data to other terminals.

[0060] The present invention also provides a computer storage medium, wherein the computer storage medium may store a program, and when the program is executed, the program may include some or all of the steps in each embodiment provided by the present invention. The storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

[0061] Therefore, the present invention provides a station area video surveillance patrol method, system, terminal and storage medium. The technical effects that can be achieved by this embodiment can be found in the description above and will not be repeated here.

[0062] Those skilled in the art can clearly understand that the technology in the embodiments of the present invention can be implemented by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution in the embodiments of the present invention, in essence or in other words, the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, and other media that can store program codes, including several instructions for enabling a computer terminal (which can be a personal computer, a server, or a second terminal, a network terminal, etc.) to execute all or part of the steps of the methods described in each embodiment of the present invention.

[0063] In this specification, the same or similar parts between the various embodiments can be referred to each other. In particular, for the terminal embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description in the method embodiment.

[0064] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or modules, which can be electrical, mechanical or other forms.

[0065] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0066] In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0067] Although the present invention has been described in detail with reference to the accompanying drawings and in combination with the preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, a person of ordinary skill in the art may make various equivalent modifications or substitutions to the embodiments of the present invention, and these modifications or substitutions shall be within the scope of the present invention. Any person of ordinary skill in the art may easily think of changes or substitutions within the technical scope disclosed by the present invention, and these shall be within the scope of protection of the present invention.

Claims

1. A station area video surveillance patrol method, characterized in that: The following steps are involved: S1, based on the patrol needs of the station area, preset patrol areas and key areas within the patrol areas, and collect basic data and target image data within the patrol areas in real time; S2, based on the radar detection algorithm combined with the target basic data, determines whether the target falls into the key area. If the target is determined to fall into the key area, the target is marked as a suspected target; S3, when the target falls into the key area, the target is tracked based on the data fusion algorithm combined with the target basic data and target image data to obtain real-time target trajectory information; Perform target detection on target image data based on convolutional neural network algorithm and obtain target clothing feature information; S4, matching the clothing feature information with a preset sample library, judging whether the target has wandering behavior based on the target trajectory information, triggering an alarm mechanism when the target's clothing feature information does not match and / or has wandering behavior, and recording the alarm information.

2. The station area video surveillance patrol method according to claim 1 is characterized in that: In step S3, the convolutional neural network algorithm specifically includes: a backbone network, a neck network and a head network; In the backbone network, the scattered feature information is integrated based on the long-range attention mechanism. At the same time, a 3×3 convolution kernel is used in the first convolution layer, and parallel plus serial max pooling is adopted. The neck network adopts a decoupled head design, which processes target recognition and target feature extraction through independent branches; The head network uses the Anchor Free mode to directly predict the candidate regions of the target. For each predicted target candidate region, its classification score and regression score are calculated, and the correct target candidate region is determined based on the classification score and regression score. The weighted comprehensive BCE Loss, Distribution Focal Loss and CIoU Loss are used as the loss function of the convolutional neural network algorithm. The loss value is transmitted back to the convolutional neural network algorithm through the back propagation algorithm and the parameters are updated.

3. The station area video surveillance patrol method according to claim 1, characterized in that: In step S3, obtaining the target's clothing feature information includes: extracting the target's upper body clothing area using an image segmentation algorithm, specifically including: Based on the top-down method, a convolutional neural network is used to obtain high-level semantic features from image data to preliminarily locate the upper body clothing area; the high-level semantic features are gradually mapped back to the original image resolution, and upsampling and convolution operations are used to accurately adjust the clothing area boundaries; The original pixel-level features of the image data are obtained based on the bottom-up method, and the local features and boundary information of the target are obtained by using clustering and edge detection algorithms to construct the upper body clothing area candidates; The upper body clothing area obtained by the Top-down method and the upper body clothing area obtained by the Bottom-up method are merged to obtain the final upper body clothing area.

4. The station area video surveillance patrol method according to claim 3 is characterized in that: In the process of fusing the upper body clothing area obtained by the Top-down method and the upper body clothing area obtained by the Bottom-up method, the following steps are specifically included: For the areas that are consistent between the two results, they are directly determined as the reliable parts of the upper body clothing area; For areas with differences, check the correlation between the difference area and the surrounding reliable parts. If the difference area should belong to the same whole as an adjacent reliable part, merge the difference area. If the difference area does not belong to the same whole as an adjacent reliable part, mark it separately.

5. The station area video surveillance patrol method according to claim 4 is characterized in that: The processing of the areas with differences also includes correction using human posture and prior knowledge: Based on the human posture estimation results and the prior knowledge of human body structure, the relationship between the difference area and various parts of the human body is evaluated; If the difference area is located near the human body's motion joints and its segmentation difference affects the rationality of the human body posture, the area is adjusted according to the deformation law of clothing under the normal motion posture of the human body.

6. The station area video surveillance patrol method according to claim 1, characterized in that: In step S4, judging whether the target has wandering behavior based on the target trajectory information specifically includes: Based on the target trajectory information, the target's trajectory length, number of movement direction changes, and residence time in a certain area within a preset time are calculated; If the length of the pedestrian's trajectory is lower than the preset length threshold, the direction of movement changes multiple times, and the time spent in the key area exceeds the preset actual threshold, the pedestrian is judged to be wandering.

7. The station area video surveillance patrol method according to claim 1, characterized in that: In step S4, the alarm information includes the number of alarms, the type of alarms and the places where the alarms often occur. The number of alarms, the type of alarms and the places where the alarms often occur are formed into a report and stored according to the preset upload interval.

8. A station area video surveillance patrol system, characterized in that: When executed, the system implements the station area video surveillance patrol method as described in claims 1-7, and the system includes: The data acquisition module presets patrol areas and key areas in patrol areas based on the patrol needs of the station area, and collects basic data and image data of targets in the patrol areas in real time; The radar judgment module determines whether the target falls into the key area based on the radar detection algorithm and the target basic data. If the target falls into the key area, the target is marked as a suspected target; Target tracking and feature extraction module: When a target falls into the key area, it tracks the target based on the data fusion algorithm and combines the target basic data and target image data to obtain real-time target trajectory information; Perform target detection on target image data based on convolutional neural network algorithm and obtain target clothing feature information; The alarm module matches the clothing feature information with the preset sample library, and determines whether the target has wandering behavior based on the target trajectory information. When the target's clothing feature information does not match and / or there is wandering behavior, the alarm mechanism is triggered and the alarm information is recorded.

9. A terminal, characterized in that: include: Memory, used to store station area video surveillance patrol programs; A processor, used to implement the steps of the station area video surveillance patrol method as described in any one of claims 1 to 7 when executing the station area video surveillance patrol program.

10. A computer-readable storage medium storing a computer program, characterized in that: The readable storage medium stores a station area video surveillance patrol program, and when the station area video surveillance patrol program is executed by the processor, the steps of the station area video surveillance patrol method according to any one of claims 1 to 7 are implemented.

Images