A shooting training safety system and method with real-time monitoring and abnormality warning

Abstract

The present invention relates to the field of anomaly monitoring technology, specifically a shooting training safety system and method for real-time monitoring and anomaly warning, the system comprising: a boundary demarcation module, a position detection module, an action construction module, an anomaly recognition module, and a signal output module. In the present invention, by extracting the three-dimensional coordinates of multiple reference points and calculating the spatial angle, a point segment boundary structure with consistent direction is constructed to achieve accurate physical identification of the training ground boundary. Based on the position changes of the shoulders, elbows, and wrists in continuous frames, velocity vectors and trajectory sequences are generated, direction changes and action trends are obtained, high-density action feature capture is achieved, and the continuous direction change segments in the trajectory are analyzed to accurately distinguish specific action types. Combined with the trajectory offset trend and boundary information, out-of-bounds behavior is identified and predicted in real time. The above processing improves the organizational level and dynamic change response capability of spatial behavior data, and achieves rapid perception and judgment of out-of-bounds risks.

Description

Technical Field

[0001] The present invention relates to the technical field of abnormality monitoring, and in particular to a shooting training safety system and method for real-time monitoring and abnormality early warning. Background Art

[0002] The field of anomaly monitoring technology involves real-time data collection and analysis of equipment operating status, environmental parameters, and operational behavior. Its core focus is on establishing effective state recognition models and risk warning mechanisms. This field encompasses three technical layers: sensor network deployment, data transmission protocol optimization, and pattern recognition algorithm design. This complete technology chain encompasses scenarios such as industrial equipment monitoring, traffic behavior management, and medical operation standards. It focuses on addressing common technical challenges such as unstructured data processing, multi-source information fusion, and low-latency response, while simultaneously meeting data collection accuracy, system robustness, and computing resource constraints.

[0003] Among them, traditional shooting training safety management refers to the process of supervising and ensuring the safety of training personnel's actions and behaviors, firearm usage status, and changes in the shooting range environment in closed or open shooting training venues through manual observation or fixed video surveillance equipment. It relies on instructors or managers to judge the training safety status based on experience, and use basic video surveillance methods to obtain image information to complete basic recording and abnormality investigation of the training process. The main means used include deploying fixed-position cameras to cover the scene, using intercom equipment for real-time communication, and relying on manual inspections to confirm whether the training personnel's actions are standardized, whether the muzzle direction is safe, and whether there are safety risks such as crossing the line or misoperation.

[0004] Existing technologies face the challenges of sparse point distribution and fuzzy boundaries when processing unstructured spatial data. This makes it difficult to accurately delineate training ground boundaries, making it difficult to construct a closed spatial reference system for dynamic behavior recognition. Consequently, out-of-bounds behaviors often go unrecognized. Furthermore, existing systems primarily rely on simple position changes to process human motion trajectories, failing to fully utilize dynamic parameters such as trajectory direction changes and velocity vectors. This leads to misjudgments and omissions of complex motion behaviors, particularly during high-speed or rapid turns. Furthermore, due to incomplete extraction of key nodes for motion recognition and reliance on a single sensor information source, posture analysis results are unstable, further limiting the depth of behavioral event recognition. Inadequate precision in boundary determination, trajectory recognition, and posture analysis impacts the reliability and timeliness of safety control, increasing the difficulty of handling abnormal events. In training scenarios, an initial movement of a person straying from a safe zone, unrecognized, can escalate into a high-risk event within seconds. However, the delayed response mechanism under traditional technology architectures cannot meet the control requirements of high-risk scenarios. Summary of the Invention

[0005] In order to solve the technical problems existing in the prior art, the embodiments of the present invention provide a shooting training safety system and method with real-time monitoring and abnormality warning. The technical solution is as follows:

[0006] On the one hand, a shooting training safety system with real-time monitoring and abnormality warning is provided, which includes:

[0007] The boundary demarcation module uses millimeter-wave radar and infrared positioning equipment to obtain point cloud data within the training ground in real time, extract the three-dimensional coordinates of multiple reference points, calculate spatial distances and angles, combine points with consistent directions to form a boundary structure, output the training ground boundary point set, and pass it to the position detection module;

[0008] The position detection module calls the training field boundary point set, uses point cloud data to detect the position of the person, obtains the coordinates of the shoulder, elbow, and wrist, calculates the distance and time difference between adjacent frames, generates a velocity vector, constructs a trajectory sequence, outputs an action trajectory feature set, and passes it to the action construction module;

[0009] The action construction module obtains the action trajectory feature set, analyzes the direction of the shoulder, elbow, and wrist trajectories, calculates the angle change between adjacent frames, extracts the trajectory segments with continuous direction changes, identifies the action type and outputs a label, outputs the posture recognition record and passes it to the anomaly recognition module;

[0010] The abnormality recognition module extracts shoulder coordinates based on the posture recognition record, identifies the movement trajectory of the person, analyzes the direction of speed change and the degree of continuous path deviation, combines boundary points, identifies and predicts cross-border behavior, outputs the cross-border event sequence and transmits it to the signal output module.

[0011] As a further solution of the present invention, the training ground boundary point set includes a boundary point coordinate set, a boundary segment direction set, and a boundary contour structure; the motion trajectory feature set includes a three-dimensional trajectory line, a velocity vector sequence, and a boundary approach mark; the posture recognition record includes an action label, a trajectory direction sequence, and a posture change pattern; and the cross-border event sequence includes a trajectory number, a cross-border starting point, and offset duration information.

[0012] As a further solution of the present invention, the boundary delineation module includes:

[0013] The data acquisition submodule uses millimeter-wave radar and infrared positioning equipment to acquire point cloud data from the training field in real time. Combining time synchronization and spatial registration, it performs data fusion and identifies point clusters in each frame where the variation in reflected signal intensity within a continuous time period is less than the set floating interval. Based on the device location tags, it selects point clusters within the observation area and generates frame-by-frame point cluster data.

[0014] The setting of the floating interval refers to setting an upper and lower limit range for the signal strength change of the same spatial point in continuous sampling frames;

[0015] The coordinate calculation submodule selects multiple reference points with continuous positions in each frame based on the amount of frame point cluster data, constructs a three-dimensional coordinate vector group according to the device positioning number, performs coordinate system conversion according to the ICP algorithm, calculates the spatial distance value between the point pairs and the angle value of the three-point combination, filters and eliminates point groups that do not meet the continuous angle change interval, and obtains the direction stable coordinate value interval;

[0016] The ICP algorithm realizes the coordinate system conversion and stable direction interval screening between point clusters by setting the point pair matching distance threshold, error convergence standard and initial positioning number;

[0017] The angle continuous change interval refers to the fluctuation range allowed for the change value of the vector angle formed by multiple reference points between consecutive frames in three-dimensional space;

[0018] The boundary generation submodule selects point segments with a direction change rate lower than the direction consistency threshold according to the directional stable coordinate value interval and performs structural combination, sorts the paths according to the direction consistency of the point segments, extracts the path vertices according to the Delaunay triangulation algorithm to construct the boundary structure, and generates a set of training field boundary points;

[0019] The direction consistency threshold is set by statistically analyzing the distribution characteristics of the direction change rate in the direction stable coordinate interval;

[0020] The Delaunay triangulation algorithm constructs a boundary vertex set based on point segments with low direction change rate, and generates a continuous and consistent training field boundary structure through the legitimacy of the circumscribed circle and boundary constraints.

[0021] As a further solution of the present invention, the position detection module includes:

[0022] The posture coordinate extraction submodule calls the three-dimensional coordinates of the boundary points in the training field boundary point set, uses the millimeter wave radar and infrared positioning equipment to detect the presence of people within the boundary range, and uses the real-time human body point cloud data collected to identify the three-dimensional spatial coordinates of the person's shoulder, elbow, and wrist in each frame. The coordinates are integrated into a continuous frame coordinate sequence according to the time sequence to generate a three-dimensional posture coordinate sequence;

[0023] A velocity trajectory calculation submodule calculates the distance between adjacent frames and the corresponding time interval based on the continuous coordinate points of the shoulder, elbow, and wrist in the three-dimensional posture coordinate sequence, constructs a velocity vector, and organizes it into a trajectory velocity sequence in chronological order;

[0024] The action feature generation submodule extracts the motion direction based on the trajectory speed sequence and the speed change trend of each segment, analyzes the direction vector changes of the shoulder, elbow, and wrist in consecutive frames, calculates the angle offset range, identifies the continuous change segments of the trajectory and organizes them into a behavioral feature sequence, and outputs the action trajectory feature set.

[0025] As a further solution of the present invention, the action building module includes:

[0026] The trajectory extraction submodule obtains the position information of the shoulder, elbow, and wrist in the action trajectory feature set, organizes the trajectory data by time series, selects continuous signal frame segments and uses them as action intervals, calculates the direction vector of the line connecting the three points in each frame based on the coordinate changes and performs normalization processing, establishes a trajectory vector set in each frame, and generates a sequence value of the part vector change;

[0027] The continuous signal frame segment refers to a frame sequence in which the coordinate data of the shoulder, elbow, and wrist are not missing or interrupted in multiple adjacent frames in the time sequence;

[0028] The direction analysis submodule calculates the vector angle between three points in adjacent frames based on the sequence of position vector changes, uses cosine similarity to obtain the angle change amplitude between frames, selects frames with an angle change fluctuation rate lower than the direction stability threshold, extracts the time segment of continuous change within the frame segment, and obtains the direction stable trajectory interval;

[0029] The specific formula for obtaining the angle variation between frames using cosine similarity is:

[0030] ;

[0031] Calculate the angle disturbance fusion value;

[0032] in, represents the angle perturbation fusion value of the i-th frame, Represents the dot product value of the shoulder-elbow direction vector and the elbow-wrist direction vector of the i-th frame, represents the product of the two vector moduli above, Indicates the displacement distance of the shoulder coordinates from the i-th frame to the i+1-th frame in three-dimensional space, represents the difference in the elbow-wrist vector angle between the i-th frame and the i+1-th frame, is the displacement disturbance adjustment factor, is the trend adjustment factor, It indicates the number of frames with the same continuous angle change trend, and i indicates the number of the currently processed frame in the time series;

[0033] The angle change fluctuation rate is obtained by calculating the standard deviation of the vector angle difference sequence between adjacent frames;

[0034] The directional stability threshold is obtained by collecting multiple stable motion trajectories, calculating the standard deviation distribution of the angle change fluctuation rate, and selecting the high percentile value as the threshold;

[0035] The action recognition submodule analyzes the angle combination pattern between the shoulder, elbow, and wrist vectors in the trajectory segment according to the directional stable trajectory interval, identifies the action type of the person in the training field, outputs the action type label, detects abnormal action types, and generates a posture recognition record sequence.

[0036] As a further solution of the present invention, the abnormality identification module includes:

[0037] A movement trajectory extraction submodule extracts the three-dimensional coordinates of the person's shoulder in consecutive frames based on the posture recognition record, arranges the coordinate points in chronological order, identifies the spatial position change trend between consecutive coordinates, generates the movement trajectory of the shoulder in the training area, and generates a movement trajectory sequence;

[0038] A path analysis submodule arranges coordinate points in chronological order according to the movement trajectory sequence, constructs a trajectory direction sequence based on the coordinate change trend, calculates the angle change between adjacent direction vectors, extracts trajectory segments with continuous angle change, extends the trajectory segments based on the end direction vectors to generate a predicted movement path sequence, and generates a path offset feature segment set;

[0039] The boundary crossing prediction submodule compares the spatial position relationship between the path and the boundary based on the path offset feature segment set, combines the movement trajectory and the boundary point set, screens the path segments that overlap with the boundary, extracts the personnel number and time position corresponding to the overlapping segments, and generates a boundary crossing event sequence.

[0040] As a further solution of the present invention, the specific formula for calculating the angle change between adjacent direction vectors is:

[0041] ;

[0042] Calculate the angle between adjacent directions;

[0043] in, Represents the direction angle between the jth frame and the j+1th frame, Represents the dot product result of the direction vector of the j-th frame and the j+1-th frame, represents the adjustment coefficient of the direction difference of the j-th frame, Represents the position offset between the jth frame and the j+1th frame, Represents the product of the direction vector modulus of the j-th frame and the j+1-th frame, represents the smoothing factor in the direction of the j-th frame, Represents the index of the current frame.

[0044] As a further embodiment of the present invention:

[0045] The signal output module obtains the corresponding number and time position in the boundary crossing event sequence, locates the abnormal path using the boundary point cloud set, outputs the warning content based on the abnormal action type recognition result, sends the warning prompt information and outputs the sound and light warning equipment control instructions to generate the abnormal behavior response record;

[0046] The control instructions of the sound and light warning device are to set sound and light alarms of different frequencies according to the length of the crossing path;

[0047] The real-time abnormal behavior recording unit includes a personnel number, an abnormal area identifier, and an alert response method.

[0048] As a further solution of the present invention, the signal output module includes:

[0049] The event matching submodule obtains the corresponding number and time position in the cross-border event sequence, matches the personnel identification corresponding to the number, extracts the abnormal action type recognition result, organizes the behavior time sequence of the number and the action tag content, and generates the event state synchronization value;

[0050] The path positioning submodule, based on the event state synchronization value, calls the spatial coordinate interval corresponding to the event time period in the boundary point cloud set, locates the spatial movement path of the person identifier within the corresponding time period, screens the coordinate point segments where the boundary crosses in the path, extracts the continuous spatial displacement trajectory, and obtains the cross-border path position sequence;

[0051] The control instruction submodule extracts the corresponding warning level type information based on the out-of-bounds path position sequence and the relationship between event time and trajectory number, matches the sound and light equipment control parameters according to the warning level type, and generates an abnormal behavior response record.

[0052] On the other hand, a shooting training safety method with real-time monitoring and abnormality early warning is provided. The method is applied to a shooting training safety system with real-time monitoring and abnormality early warning. The method includes:

[0053] S1: Using millimeter-wave radar and infrared positioning equipment, we acquire point cloud data from the training ground in real time, extract the 3D coordinates of multiple reference points, calculate the distances and directional angles between these points based on the point cloud spatial coordinates, classify point segments with consistent directional changes, integrate them into continuous boundary line segments, and output them as a set of training ground boundary points.

[0054] S2: Obtain the boundary point set of the training ground, use point cloud data to detect the three-dimensional coordinate data of the person in the training ground, obtain the space of the shoulder, elbow, and wrist, calculate the distance and time difference between adjacent time frames, generate a velocity vector, construct a point-continuous trajectory path according to the time series, and output it as a motion trajectory feature set;

[0055] S3: Obtain the motion trajectory feature set, extract the direction angle change value between each pair of adjacent frames in the shoulder, elbow, and wrist trajectory vectors, extract the trajectory segment with continuous direction change, identify the motion type and output the corresponding motion label, and output it as a posture recognition record;

[0056] S4: Obtain the posture recognition record, extract the shoulder coordinate sequence, identify the movement trajectory of the person, analyze the speed change direction and the degree of continuous path deviation, combine the boundary point position, identify and predict the cross-border behavior, and output the cross-border event sequence;

[0057] S5: Obtain the corresponding number and time position of the cross-border event sequence, use the boundary point cloud set to locate the abnormal path, combine the action label content in the corresponding posture recognition record, generate warning content according to the time index and abnormal path number, send warning prompt information and output control instructions for the audio-visual equipment, and generate an abnormal behavior response record.

[0058] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0059] By acquiring point cloud data from the training ground and extracting the three-dimensional coordinates of multiple reference points, combined with spatial distance and angle calculations, the system combines points with consistent directions to form a three-dimensional boundary structure. This effectively improves the accuracy of determining the physical enclosure of complex spatial scenes and enhances the accuracy of environmental modeling. Furthermore, it uses point cloud data from consecutive frames to locate key nodes such as the shoulder, elbow, and wrist. Velocity vectors are calculated using inter-frame distance and time difference, thereby constructing a dynamic feature sequence that includes directional changes and trajectory extensions. This allows for more timely and granular expression of motion trajectories. By analyzing the angle changes between adjacent frames in the trajectory and extracting trajectory segments with continuous directions, the system can efficiently identify the action type and record the posture, providing structured input for subsequent judgment. Furthermore, by combining the identified trajectory with action posture information, it extracts shoulder coordinate change trends and continuous trajectory offset features. This information is then integrated with boundary structure information to identify and predict boundary violations, enabling real-time prediction and timely response to boundary violations. In the overall processing logic, the coordinated cooperation of multi-point three-dimensional information extraction, multi-level trajectory analysis and high-frequency posture change recognition significantly enhances the fine-grained capture capability of spatial motion behaviors, improves the detection accuracy and response efficiency of cross-border events, and realizes the accurate identification and risk prevention of dynamic behaviors within the safety control boundary. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0061] Figure 1 is a system flow chart of the present invention;

[0062] Figure 2 Schematic diagram of the system framework of the present invention;

[0063] Figure 3 Schematic diagram of the method steps of the present invention. DETAILED DESCRIPTION

[0064] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0065] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0066] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same. The terms "of," "corresponding," and "corresponding" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same.

[0067] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0068] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0069] The embodiment of the present invention provides a shooting training safety system with real-time monitoring and abnormal warning. Figures 1 to 2 The present invention provides a technical solution, a shooting training safety system with real-time monitoring and abnormal warning, comprising:

[0070] The boundary demarcation module uses millimeter-wave radar and infrared positioning equipment to obtain point cloud data within the training ground in real time, extract the three-dimensional coordinates of multiple reference points, calculate spatial distances and angles, combine points with consistent directions to form a boundary structure, output the training ground boundary point set, and pass it to the position detection module;

[0071] The position detection module calls the training field boundary point set, uses point cloud data to detect the person's position, obtains the coordinates of the shoulder, elbow, and wrist, calculates the distance and time difference between adjacent frames, generates a velocity vector, constructs a trajectory sequence, outputs the action trajectory feature set, and passes it to the action construction module;

[0072] The action construction module obtains the action trajectory feature set, analyzes the trajectory directions of the shoulder, elbow, and wrist, calculates the angle change between adjacent frames, extracts the trajectory segments with continuous direction changes, identifies the action type and outputs the label, outputs the posture recognition record and passes it to the anomaly recognition module;

[0073] The anomaly recognition module extracts shoulder coordinates based on posture recognition records, identifies the movement trajectory of the person, analyzes the direction of speed change and the degree of continuous path deviation, combines boundary points, identifies and predicts cross-border behavior, outputs the cross-border event sequence, and transmits it to the signal output module;

[0074] The signal output module obtains the corresponding number and time position in the boundary crossing event sequence, locates the abnormal path using the boundary point cloud set, and outputs the warning content based on the abnormal action type recognition results, sends warning prompt information and outputs the sound and light warning equipment control instructions to generate abnormal behavior response records;

[0075] The control instructions of the sound and light warning equipment are to set sound and light alarms of different frequencies according to the length of the cross-border path.

[0076] The training ground boundary point set includes the boundary point coordinate set, the boundary segment direction set, and the boundary contour structure. The action trajectory feature set includes the three-dimensional trajectory line, the velocity vector sequence, and the boundary approach mark. The posture recognition record includes the action label, the trajectory direction sequence, and the posture change pattern. The cross-border event sequence includes the trajectory number, the cross-border starting point, and the offset duration information. The real-time abnormal behavior recording unit includes the personnel number, the abnormal area identification, and the warning response method.

[0077] See also Figure 1 and Figure 2 , the boundary delineation module includes:

[0078] The data acquisition submodule uses millimeter-wave radar and infrared positioning equipment to acquire point cloud data from the training field in real time. Combining time synchronization and spatial registration, it performs data fusion and identifies point clusters in each frame where the variation in reflected signal intensity within a continuous time period is less than the set floating interval. Based on the device location tags, it selects point clusters within the observation area and generates frame-by-frame point cluster data.

[0079] First, point cloud data from the training ground is acquired using millimeter-wave radar and infrared positioning devices. The devices' real-time sampling rate is set to 100 frames per second. Each frame of point cloud data includes reflected signal strength, a timestamp, and a location tag. To ensure high data accuracy, device time synchronization accuracy is required to be within the microsecond level. After synchronization, the data is spatially aligned. This involves merging the data collected by different devices into a point cloud in a unified coordinate system using a precise spatial transformation algorithm based on the device's positioning information. This spatial alignment method uses the linear least squares method for optimal alignment, ensuring high accuracy of the spatial position of each point in the merged point cloud. Next, for each frame of data, point clusters whose reflected signal strength changes within a set floating range are identified. This floating range is determined based on historical data statistics and is set to a ±5% signal strength fluctuation range. For example, if the signal strength at a specific acquisition point is 100dB, the floating range is between 95dB and 105dB. If the reflected signal strength of a point cluster within a certain time period changes within this range, the cluster is considered eligible. On this basis, the device location tag will be used to filter the point clusters located within the observation area. For example, assuming the device location tag is (x=100, y=200, z=50), if the spatial location of a point cluster falls within the set observation area, the point cluster is considered to meet the requirements. After this screening process, the final frame point cluster data volume is generated for subsequent processing;

[0080] Setting a floating interval means setting an upper and lower limit range for the signal strength change at the same spatial point in continuous sampling frames;

[0081] The coordinate calculation submodule selects multiple reference points with continuous positions in each frame based on the amount of frame point cluster data, constructs a three-dimensional coordinate vector group according to the device positioning number, performs coordinate system transformation based on the ICP algorithm, calculates the spatial distance between point pairs and the angle between three points, and screens and eliminates point groups that do not meet the continuous angle change range to obtain the directionally stable coordinate value range;

[0082] In the coordinate calculation submodule, multiple reference points with continuous positions in each frame of data are first selected as reference coordinate points. Suppose three reference points are selected, namely A (10, 20, 30), B (15, 25, 35), and C (20, 30, 40). The spatial positions of these points are obtained by infrared positioning equipment. Then, based on the data of these reference points, a three-dimensional coordinate vector group is constructed according to the positioning number of the device, and the ICP algorithm (iterative closest point algorithm) is used for coordinate system conversion. The ICP algorithm calculates the distance between reference points A, B, and C to select point pairs with smaller distance errors. The specific calculation process is to first calculate the spatial distance between the point pairs, for example, to calculate the distance from point A to point B, that is: , substituting the coordinates of A(10,20,30) and B(15,25,35), we get:

[0083] ;

[0084] If the error of the point pair meets the set threshold (for example, the threshold is set to 10 units), the point pair is considered valid. Based on these valid point pairs, the spatial distance between the point pairs and the angle between the three points are calculated. For three points A, B, and C, the angle between the two vectors AB and BC formed by them is calculated. For example, the vector dot product formula can be used to calculate the angle: , further screening and eliminating point groups that do not conform to the continuous angle change range. The continuous angle change range is obtained by statistically analyzing the angle change range between reference points in multiple consecutive frames. Assuming that the set angle change range is ±10°, if the angle change between consecutive frames exceeds this range, the point group is eliminated, and the final direction stable coordinate value range is obtained;

[0085] The ICP algorithm realizes the coordinate system conversion and stable direction interval screening between point clusters by setting the point pair matching distance threshold, error convergence standard and initial positioning number;

[0086] The angle continuous change interval refers to the allowed fluctuation range of the vector angle change value composed of multiple reference points between consecutive frames in three-dimensional space.

[0087] The boundary generation submodule selects points with a direction change rate lower than the direction consistency threshold according to the direction stable coordinate value interval and performs structural combination. It sorts the paths according to the direction consistency of the points and segments, extracts the path vertices according to the Delaunay triangulation algorithm to construct the boundary structure, and generates a set of training field boundary points.

[0088] In the boundary generation submodule, first, based on the direction-stable coordinate value interval, point segments with a direction change rate lower than the direction consistency threshold are selected for structural combination. Suppose in a certain frame of data, point segments AB and BC are selected as candidate point segments, and their direction change rates are 2° and 5° respectively, and the set direction consistency threshold is 10°. Since the direction change rates of both are lower than the threshold, these two point segments can be structurally combined. Next, the path is sorted according to the direction consistency of the point segments to obtain the sorting results of each point segment. Taking point segments AB and BC as an example, after sorting based on direction consistency, a new point segment order is obtained, and the Delaunay triangulation algorithm is used to extract the path vertices to construct the boundary structure. Suppose points A (10, 20, 30), B (15, 25, 35), and C (20, 30, 40) are selected from the corners of the training ground in turn. According to the Delaunay algorithm, the lines connecting the three points are extracted to form triangles to generate the boundary structure. Finally, a set of boundary points of the training ground is obtained, which represent the actual boundary position of the training ground;

[0089] The direction consistency threshold is set by statistically analyzing the distribution characteristics of the direction change rate in the direction stable coordinate interval;

[0090] The Delaunay triangulation algorithm constructs a set of boundary vertices based on point segments with low directional change rates, and generates a continuous and consistent training ground boundary structure through the legality of the circumscribed circle and boundary constraints;

[0091] See also Figure 1 and Figure 2 , the position detection module includes:

[0092] The posture coordinate extraction submodule calls the 3D coordinates of the boundary points in the training field boundary point set, uses millimeter-wave radar and infrared positioning equipment to detect the presence of people within the boundary range, and uses the real-time human body point cloud data to identify the 3D spatial coordinates of the person's shoulder, elbow, and wrist in each frame. It integrates the data into a continuous frame coordinate sequence according to the time sequence to generate a 3D posture coordinate sequence.

[0093] First, through millimeter-wave radar and infrared positioning equipment, point cloud data in the training ground is obtained in real time. These data will include the three-dimensional coordinates of the reference points in the training ground, and the spatial layout of the boundary at each moment is determined based on this. By analyzing the three-dimensional coordinates of the boundary points and combining the position of the personnel within the boundary range, the coordinates of the personnel's shoulders, elbows, wrists and other key positions are extracted in real time using point cloud data. These point cloud data will be calibrated at each acquisition moment, and the three-dimensional spatial position of the shoulders, elbows and wrists will be identified. For example, suppose that at a certain moment, the coordinates of the shoulder are , elbow is , wrist is , the system will integrate these data to form a set of continuous three-dimensional coordinate sequences. Based on the time sequence, these frame data will be associated one by one to generate an overall three-dimensional posture coordinate sequence. By synthesizing the coordinates of consecutive frames, it is possible to accurately describe the movement trajectory of each key point over time and ultimately output a three-dimensional posture coordinate sequence containing multiple frames of data. If the coordinates of the current time point are , then the posture sequence of consecutive frames is , providing data for further motion analysis.

[0094] The velocity trajectory calculation submodule calculates the distance between adjacent frames and the corresponding time interval based on the continuous coordinate points of the shoulder, elbow, and wrist in the 3D posture coordinate sequence, constructs the velocity vector, and organizes it into a trajectory velocity sequence in chronological order;

[0095] The core task of the module is to calculate the motion speed between adjacent frames based on the continuous shoulder, elbow, and wrist position data extracted from the posture coordinate sequence. First, the three-dimensional coordinate values ​​of the shoulder, elbow, and wrist are obtained at each moment, and then the coordinate changes between adjacent frames are calculated. For example, for the shoulder in frame and The calculation process of the displacement between , and then according to the time interval , velocity vector It will be calculated by the following formula: In the same way, the velocity vectors of the elbow and wrist will also be calculated, and the corresponding velocity trajectory sequences will be constructed. These velocity vectors will be sorted in chronological order for subsequent behavioral analysis. and The shoulder displacement is meters, the time interval is seconds, the calculated velocity vector is: m / s, so it can be obtained that the velocity trajectory sequence will include the velocity values ​​of multiple frames, constituting the velocity evolution information of the entire motion process for use in the next stage of motion analysis.

[0096] The motion feature generation submodule extracts the motion direction based on the trajectory velocity sequence and the velocity change trend of each segment. It analyzes the changes in the direction vectors of the shoulder, elbow, and wrist in consecutive frames, calculates the angle offset range, identifies the continuous change segments of the trajectory, organizes them into a behavioral feature sequence, and outputs the motion trajectory feature set.

[0097] Based on the constructed trajectory speed sequence, the trend of speed change in each segment will be further analyzed. First, the motion direction in each trajectory segment is extracted, the direction vector changes of the shoulder, elbow, and wrist between frames are analyzed, and the angle offset range between direction vectors is calculated. For each segment of motion, the system will calculate the angle change between consecutive frames. For example, if at two time points and The shoulder direction vector is

[0098] , and in Chuwei , then the angle It will be calculated by the following formula:

[0099] , in this case, the angle is calculated to be , the change in the direction vector at this point indicates a significant change in direction. Through this type of calculation, the module can extract trajectory change segments from continuous coordinate and velocity data and identify specific action types based on these segments. Ultimately, these extracted continuous trajectory change segments are organized into behavioral feature sequences and output as a set of action trajectory features for subsequent anomaly detection and behavioral analysis.

[0100] See also Figure 1 and Figure 2 , the action building blocks include:

[0101] The trajectory extraction submodule obtains the position information of the shoulder, elbow, and wrist in the action trajectory feature set, organizes the trajectory data by time series, selects continuous signal frames and uses them as action intervals, calculates the direction vector of the line connecting the three points in each frame based on the coordinate changes and performs normalization processing, establishes a trajectory vector set in each frame, and generates a sequence value of the part vector change;

[0102] First, the position information of the shoulder, elbow, and wrist is obtained, and these coordinate data are sorted in chronological order to generate trajectory data of continuous frames. The coordinates of the shoulder, elbow, and wrist at each time point will be sorted into a time series in turn. After screening, these data can provide a continuous frame segment of the signal. The definition of a continuous frame segment of the signal is that in the time series, the coordinate data of the shoulder, elbow, and wrist are not missing or interrupted in multiple adjacent frames. Taking a simple example as an example, suppose that at a certain moment, the shoulder position is , the elbow position is , the wrist position is , these data are used to calculate the coordinate changes and obtain the direction vector in each frame and Next, the vectors of each frame are normalized to obtain a normalized direction vector, and these direction vectors are combined into a set of trajectory vectors. Finally, a sequence of vector changes for each part is generated by combining all these trajectory vectors, and this data is used as input for subsequent motion analysis.

[0103] A continuous signal frame segment refers to a frame sequence in which the coordinate data of the shoulder, elbow, and wrist are not missing or interrupted in multiple adjacent frames in the time series.

[0104] The direction analysis submodule calculates the vector angle between three points in adjacent frames based on the sequence of part vector changes. It uses cosine similarity to obtain the angle change amplitude between frames, selects frames where the angle change fluctuation rate is lower than the direction stability threshold, extracts the time segments of continuous changes within the frame segments, and obtains the direction stable trajectory interval.

[0105] The specific formula for obtaining the angle variation between frames using cosine similarity is:

[0106] ;

[0107] Calculate the angle disturbance fusion value;

[0108] in, represents the angle perturbation fusion value of the i-th frame, Represents the dot product value of the shoulder-elbow direction vector and the elbow-wrist direction vector of the i-th frame, represents the product of the two vector moduli above, Indicates the displacement distance of the shoulder coordinates from the i-th frame to the i+1-th frame in three-dimensional space, represents the difference in the elbow-wrist vector angle between the i-th frame and the i+1-th frame, is the displacement disturbance adjustment factor, is the trend adjustment factor, It indicates the number of frames with the same continuous angle change trend, and i indicates the number of the currently processed frame in the time series;

[0109] formula:

[0110] ;

[0111] Detailed explanation of the formula and calculation process:

[0112] This formula is used to calculate the The frame angle perturbation fusion value takes into account factors such as the similarity between the shoulder-elbow direction vector and the elbow-wrist direction vector, shoulder displacement, elbow-wrist angle change, and trend adjustment. The following is a detailed explanation of each parameter in the formula and calculations using actual data;

[0113] Parameter meaning and setting value:

[0114] : No. Set the dot product value of the frame shoulder-elbow direction vector and the elbow-wrist direction vector. ;

[0115] : No. The product of the module length of the shoulder-elbow direction vector and the elbow-wrist direction vector, ;

[0116] : No. Frame to The displacement distance of the frame shoulder coordinates in three-dimensional space, set ;

[0117] : No. Frame and The difference in the elbow-wrist vector angle between frames, set ;

[0118] : Displacement disturbance adjustment factor, usually between 0.1 and 0.5, set ;

[0119] : Trend adjustment factor, usually between 0.01 and 0.1, set ;

[0120] : The number of frames that the forward continuous angle change trend remains consistent, usually between 2 and 10, set ;

[0121] Enter the formula for calculation:

[0122] Calculation process:

[0123] Calculate the cosine similarity part:

[0124] ;

[0125] Calculate the displacement disturbance part:

[0126] ;

[0127] ;

[0128] Calculate the trend adjustment part:

[0129] ;

[0130] Calculate the angle disturbance fusion value:

[0131] ;

[0132] This result shows that the The angle perturbation fusion value of the frame is approximately 3.772, indicating that in this frame, the similarity between the shoulder-elbow direction vector and the elbow-wrist direction vector is high, the impact of displacement perturbation and elbow-wrist angle change on the angle perturbation is moderate, and the angle change trend between consecutive frames remains consistent.

[0133] The fluctuation rate of angle change is obtained by calculating the standard deviation of the vector angle difference sequence between adjacent frames;

[0134] The directional stability threshold is obtained by collecting multiple stable motion trajectories, calculating the standard deviation distribution of the angle change fluctuation rate, and selecting the high percentile value as the threshold.

[0135] The action recognition submodule analyzes the angle combination pattern between the shoulder, elbow, and wrist vectors in the trajectory segment according to the direction-stable trajectory interval, identifies the action type of the person in the training field, outputs the action type label, detects abnormal action types, and generates a posture recognition record sequence;

[0136] In the action recognition submodule, the selected directional stable trajectory intervals are first used as the basis for analysis. These intervals have been processed by the previous submodule to ensure signal continuity and no missing. Next, the vector combination pattern between the shoulder, elbow, and wrist in each frame is analyzed. This process includes calculating the angle between each vector to determine the posture change during the movement. For example, suppose that in a certain frame, the vector from the shoulder to the elbow is , the vector from the elbow to the wrist is , then according to the angle between these two vectors Perform the calculation: Through this calculation, the action features of each frame can be obtained, and by combining these features, the specific action types of people in the training ground can be identified. If the angle change of a certain action is significantly different from the known standard posture (such as raising a gun, aiming, shooting, etc.), the system will identify it as an abnormal action. By comparing the posture and angle changes between consecutive frames, a posture recognition record sequence is generated for further abnormal action detection. According to different action types and identified abnormal actions, the module outputs the corresponding action type label and abnormal action recognition results;

[0137] Table 1 Trajectory data samples

[0138]

[0139] As shown in Table 1, the coordinates of the shoulder, elbow, and wrist in the three-dimensional position data are integrated through time series, and the direction vector is calculated. After normalization in each frame of data, the direction vector is output to further generate the motion trajectory.

[0140] See also Figure 1 and Figure 2 , the anomaly recognition module includes:

[0141] The movement trajectory extraction submodule extracts the three-dimensional coordinates of the person's shoulder in consecutive frames based on posture recognition records, arranges the coordinate points in chronological order, identifies the spatial position change trend between consecutive coordinates, generates the movement trajectory of the shoulder in the training area, and generates a movement trajectory sequence;

[0142] In the trajectory extraction submodule, first, the posture data of the personnel in the training ground are recorded through posture recognition technology, especially the three-dimensional coordinates of the shoulders. These data are collected once per second based on the sensor system, and the spatial coordinate values ​​at each time point are recorded. For example, the three-dimensional coordinates of the shoulder recorded at a certain moment are (x=5.0, y=3.2, z=1.5). This coordinate is recorded synchronously with the timestamp to form a time series. Then, the obtained coordinates of each frame are arranged in chronological order to ensure that the data sequence is correct. By calculating the spatial position change trend between consecutive coordinate points, specifically, calculating the distance change between each two adjacent coordinate points, the following formula is used: , assuming that the shoulder coordinates at a certain moment are (x1=5.0,y1=3.2,z1=1.5), and the shoulder coordinates at the next moment are (x2=5.2,y2=3.3,z2=1.6), then the spatial distance between the two points is: ,Through this method, the continuous coordinate change trend will reflect the movement ,trajectory of the shoulder. Combined with the entire time series, a complete shoulder movement ,trajectory sequence is generated. For example, if the continuous coordinate sequence is (5.0, 3.2, 1.5), ,(5.2, 3.3, 1.6), (5.5, 3.6, 1.8), then these data form a path, ,representing the movement trajectory of the person in the training area.

[0143] The path analysis submodule arranges coordinate points in chronological order according to the movement trajectory sequence, constructs a trajectory direction sequence based on the coordinate change trend, calculates the angle change between adjacent direction vectors, extracts trajectory segments with continuous angle changes, and extends them based on the direction vectors at the end points of the trajectory segments to generate a predicted movement path sequence and a set of path offset feature segments.

[0144] formula:

[0145] ;

[0146] Detailed explanation of the formula and the process of formula calculation and derivation:

[0147] The formula is used to calculate the angle between the direction vector of the jth frame and the j+1th frame, which is used to determine the direction change trend between the two frames. The result is used to extract the trajectory segments with continuous angle changes to support the generation process of the predicted movement path;

[0148] Parameter meaning and setting value:

[0149] is the dot product value of the direction vector of the j-th frame and the j+1-th frame. The dot product calculation formula is: , set the current frame vector , the next frame vector ,but ;

[0150] It is the direction difference adjustment coefficient, which ranges from 0.05 to 0.15 in stable operation and is set to 0.1 in this case.

[0151] is the position offset between the jth frame and the j+1th frame, which is calculated as the Euclidean distance of the difference between the spatial coordinates of the two frames. , set the acquisition coordinates to and ;but ;

[0152] It is the product of the two frame direction vector modulus lengths. The vector modulus length is calculated as: ;

[0153] is the directional smoothing factor. When the standard deviation is set to 0.03, the corresponding factor value is set to 0.03;

[0154] Substitute the parameters into the formula for calculation:

[0155] ;

[0156] ;

[0157] The result of 10.1° indicates that the angle change of the direction vector between the current frame and the next frame is in a relatively continuous state. The angle change value will be compared with the set angle continuous change threshold. If the angle change values ​​in consecutive frames do not exceed the threshold, the corresponding trajectory segment is extracted as an angle continuous change trajectory segment, providing a trajectory basis for subsequent movement path prediction.

[0158] The boundary crossing prediction submodule compares the spatial relationship between the path and the boundary based on the path offset feature segment set, combined with the movement trajectory and the boundary point set, screens the path segments that overlap with the boundary, extracts the person number and time position corresponding to the overlapping segments, and generates a boundary crossing event sequence;

[0159] In the boundary crossing prediction submodule, the spatial relationship between the individual's movement trajectory and the set of training ground boundary points is first extracted by identifying path deviation feature segments. A path deviation feature segment refers to a portion of the trajectory that deviates significantly from the expected path, typically with significant path changes. For example, if a trajectory in a given frame moves from (x=5.0, y=3.2) to (x=8.0, y=4.0), this significant deviation is considered a path deviation. Next, the path is compared with the boundary based on the data from the boundary point set to determine if there is any overlap. For example, if the boundary points are (x=6.0, y=4.0) and (x=8.0, y=5.0), and a path segment passes through the area between these two points, the path segment is considered to overlap with the boundary. Overlapping path segments are then filtered, and the corresponding person ID and time location are recorded. For example, if a path segment starts at (x=7.8, y=3.5) and ends at (x=8.2, y=4.1), and the path segment overlaps within the set range of the training area boundary, the selected overlapping path segment will correspond to a specific person number (such as 001) and a timestamp (such as 15:32:10). In this way, a cross-border event sequence is generated, and the cross-border information of the person is recorded, which facilitates the subsequent tracking and processing of cross-border behavior;

[0160] Table 2 Cross-border prediction result data table

[0161]

[0162] As shown in Table 2, the boundary crossing prediction submodule recorded the boundary crossing events for personnel numbers 001 and 002. Each record includes the start time, end time, starting coordinates, and ending coordinates. This information helps the system quickly locate the boundary crossing behavior and compare it with the training area boundaries, further ensuring the safety and regularity of the training ground.

[0163] See also Figure 1 and Figure 2 , the signal output module includes:

[0164] The event matching submodule obtains the corresponding number and time position in the cross-border event sequence, matches the personnel identification corresponding to the number, extracts the abnormal action type recognition result, organizes the behavior time sequence of the number and the action tag content, and generates the event state synchronization value;

[0165] In the event matching submodule, first, the detailed information of each event is obtained based on the number and time position in the cross-border event sequence. For each cross-border event, the system will match the corresponding person ID according to the event number and extract the abnormal action type identification results of the person during the event time period. The system associates the event number with the person ID and integrates the behavior time series with the action tag content to generate an event state synchronization value containing the detailed information of the event. For example, in a certain cross-border event, assuming the number is , the event time is arrive , the personnel identification is , the abnormal action type is "beyond the shooting boundary", and a synchronization value is formed through the action mark content and time data extracted by the system , as input for subsequent processing, ensuring accurate association between person identification, time location and action type for each event.

[0166] The path positioning submodule, based on the event state synchronization value, calls the spatial coordinate interval corresponding to the event time period in the boundary point cloud set, locates the spatial movement path of the person in the corresponding time period, filters the coordinate point segments where the boundary crosses in the path, extracts the continuous spatial displacement trajectory, and obtains the cross-border path position sequence;

[0167] In the path positioning submodule, based on the event state synchronization value, the system first locates the spatial coordinate interval corresponding to the event time period by calling the data in the boundary point cloud set. By tracking the movement path of the personnel during the time period, the system identifies the spatial path of the personnel during the event. These paths will be used to filter the coordinate point segments where the boundary crosses. For example, if the time period of an event is arrive , the system will filter out the coordinate data within this period from the boundary point cloud collection and mark the trajectory points where the boundary crosses. When a section of the path is detected to exceed the set boundary, this section will be extracted and marked as an out-of-bounds path. For example, in the event In the time period arrive The coordinate segment of the path between the two paths that exceeds the boundary is , this part of the path is identified as an out-of-bounds path and output as an out-of-bounds path position sequence , for use by subsequent control instruction modules.

[0168] The control instruction submodule extracts the corresponding warning level type information based on the out-of-bounds path position sequence and the relationship between event time and trajectory number, matches the sound and light equipment control parameters according to the warning level type, and generates abnormal behavior response records;

[0169] The function of the control instruction submodule is to extract the corresponding warning level type information based on the cross-border path position sequence, combined with the relationship between the event time and the trajectory number, and generate the control parameters of the sound and light equipment according to the warning level. In this module, the system first extracts the warning level type (such as "high-risk warning", "medium-risk warning", etc.) according to the event time corresponding to the cross-border path position sequence, and matches the corresponding equipment control parameters according to the warning level. For example, assuming that in the event In the example, the out-of-bounds path position sequence is According to pre-set rules, when the warning level of the cross-border path is "high-risk warning", the system will match the corresponding sound and light equipment control parameters, such as sound frequency 500Hz, light intensity 80%. Ultimately, these control parameters will generate abnormal behavior response records, record the response operation of the equipment, and provide timely warning prompts to relevant personnel. For example, when the system recognizes event number When it is a high-risk warning, the control instruction generated is "500Hz frequency, 80% light intensity".

[0170] See also Figure 3 , methods include:

[0171] S1: Using millimeter-wave radar and infrared positioning equipment, we acquire point cloud data from the training ground in real time, extract the 3D coordinates of multiple reference points, calculate the distances and directional angles between these points based on the point cloud spatial coordinates, classify point segments with consistent directional changes, integrate them into continuous boundary line segments, and output them as a set of training ground boundary points.

[0172] S2: Obtain a set of boundary points of the training ground, use point cloud data to detect the three-dimensional coordinate data of the person in the training ground, obtain the space of the shoulder, elbow, and wrist, calculate the distance and time difference between adjacent time frames, generate a velocity vector, and construct a point-continuous trajectory path according to the time series. The output is a motion trajectory feature set;

[0173] S3: Obtain the action trajectory feature set, extract the direction angle change value between each pair of adjacent frames in the shoulder, elbow, and wrist trajectory vectors, extract the trajectory segments with continuous direction changes, identify the action type and output the corresponding action label, which is output as a posture recognition record;

[0174] S4: Acquire posture recognition records, extract shoulder coordinate sequences, identify personnel movement trajectories, analyze speed change directions and path continuous deviations, combine boundary point locations, identify and predict boundary crossing behaviors, and output boundary crossing event sequences;

[0175] S5: Obtain the corresponding number and time position of the out-of-bounds event sequence, use the boundary point cloud set to locate the abnormal path, combine the action label content in the corresponding posture recognition record, generate warning content according to the time index and abnormal path number, send warning prompt information and output control instructions for the audio and video equipment, and generate abnormal behavior response records.

[0176] The above embodiments can be implemented in whole or in part via software, hardware (e.g., circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in accordance with the embodiments of the present invention are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0177] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0178] In this disclosure, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of a, b, or c" can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0179] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0180] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0181] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0182] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, 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 interface, indirect coupling or communication connection of the device or unit, which can be electrical, mechanical or other forms.

[0183] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0184] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0185] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical disks.

[0186] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A shooting training safety system with real-time monitoring and abnormal warning, characterized by: The system comprises: The boundary delineation module obtains point cloud data within the training field, extracts the three-dimensional coordinates of multiple reference points, calculates spatial distances and angles, combines points with consistent directions to form a boundary structure, outputs a set of training field boundary points, and passes it to the position detection module; The position detection module calls the training field boundary point set, uses point cloud data to detect the position of the person, obtains the coordinates of the shoulder, elbow, and wrist, calculates the distance and time difference between adjacent frames, generates a velocity vector, constructs a trajectory sequence, outputs an action trajectory feature set, and passes it to the action construction module; The action construction module obtains the action trajectory feature set, analyzes the direction of the shoulder, elbow, and wrist trajectories, calculates the angle change between adjacent frames, extracts the trajectory segments with continuous direction changes, identifies the action type and outputs a label, outputs the posture recognition record and passes it to the anomaly recognition module; An anomaly recognition module, based on the posture recognition record, extracts shoulder coordinates, identifies the movement trajectory of the person, analyzes the direction of speed change and the degree of continuous path deviation, combines boundary points, identifies and predicts cross-border behavior, outputs a cross-border event sequence and transmits it to the signal output module; The boundary delineation module includes: The data acquisition submodule uses millimeter-wave radar and infrared positioning equipment to acquire point cloud data from the training field in real time. Combining time synchronization and spatial registration, it performs data fusion and identifies point clusters in each frame where the variation in reflected signal intensity within a continuous time period is less than the set floating interval. Based on the device location tags, it selects point clusters within the observation area and generates frame-by-frame point cluster data. The setting of the floating interval refers to setting an upper and lower limit range for the signal strength change of the same spatial point in continuous sampling frames; The coordinate calculation submodule selects multiple reference points with continuous positions in each frame based on the amount of frame point cluster data, constructs a three-dimensional coordinate vector group according to the device positioning number, performs coordinate system conversion according to the ICP algorithm, calculates the spatial distance value between the point pairs and the angle value of the three-point combination, filters and eliminates point groups that do not meet the continuous angle change interval, and obtains the direction stable coordinate value interval; The ICP algorithm realizes the coordinate system conversion and stable direction interval screening between point clusters by setting the point pair matching distance threshold, error convergence standard and initial positioning number; The angle continuous change interval refers to the fluctuation range allowed for the change value of the vector angle formed by multiple reference points between consecutive frames in three-dimensional space; The boundary generation submodule selects point segments with a direction change rate lower than the direction consistency threshold according to the directional stable coordinate value interval and performs structural combination, sorts the paths according to the direction consistency of the point segments, extracts the path vertices according to the Delaunay triangulation algorithm to construct the boundary structure, and generates a set of training field boundary points; The direction consistency threshold is set by statistically analyzing the distribution characteristics of the direction change rate in the direction stable coordinate interval; The Delaunay triangulation algorithm constructs a boundary vertex set based on point segments with low direction change rate, and generates a continuous and consistent training field boundary structure through the legitimacy of the circumscribed circle and boundary constraints.

2. The shooting training safety system with real-time monitoring and abnormal warning according to claim 1 is characterized in that: The training ground boundary point set includes a boundary point coordinate set, a boundary segment direction set, and a boundary contour structure; the action trajectory feature set includes a three-dimensional trajectory line, a velocity vector sequence, and a boundary approach mark; the posture recognition record includes an action label, a trajectory direction sequence, and a posture change pattern; and the cross-border event sequence includes a trajectory number, a cross-border starting point, and offset duration information.

3. The shooting training safety system with real-time monitoring and abnormal warning according to claim 1 is characterized in that: The position detection module includes: The posture coordinate extraction submodule calls the three-dimensional coordinates of the boundary points in the training field boundary point set, uses millimeter-wave radar and infrared positioning equipment to detect the presence of people within the boundary range, and uses the real-time human body point cloud data collected to identify the three-dimensional spatial coordinates of the person's shoulder, elbow, and wrist in each frame. The coordinates are integrated into a continuous frame coordinate sequence according to the time sequence to generate a three-dimensional posture coordinate sequence; A velocity trajectory calculation submodule calculates the distance between adjacent frames and the corresponding time interval based on the continuous coordinate points of the shoulder, elbow, and wrist in the three-dimensional posture coordinate sequence, constructs a velocity vector, and organizes it into a trajectory velocity sequence in chronological order; The action feature generation submodule extracts the motion direction based on the trajectory speed sequence and the speed change trend of each segment, analyzes the direction vector changes of the shoulder, elbow, and wrist in consecutive frames, calculates the angle offset range, identifies the continuous change segments of the trajectory and organizes them into a behavioral feature sequence, and outputs the action trajectory feature set.

4. The shooting training safety system with real-time monitoring and abnormal warning according to claim 3 is characterized in that: The action building block includes: The trajectory extraction submodule obtains the position information of the shoulder, elbow, and wrist in the action trajectory feature set, organizes the trajectory data by time series, selects continuous signal frame segments and uses them as action intervals, calculates the direction vector of the line connecting the three points in each frame based on the coordinate changes and performs normalization processing, establishes a trajectory vector set in each frame, and generates a sequence value of the part vector change; The continuous signal frame segment refers to a frame sequence in which the coordinate data of the shoulder, elbow, and wrist are not missing or interrupted in multiple adjacent frames in the time sequence; The direction analysis submodule calculates the vector angle between three points in adjacent frames based on the sequence of position vector changes, uses cosine similarity to obtain the angle change amplitude between frames, selects frames with an angle change fluctuation rate lower than the direction stability threshold, extracts the time segment of continuous change within the frame segment, and obtains the direction stable trajectory interval; The specific formula for obtaining the angle variation between frames using cosine similarity is: ； Calculate the angle disturbance fusion value; in, represents the angle perturbation fusion value of the i-th frame, Represents the dot product value of the shoulder-elbow direction vector and the elbow-wrist direction vector of the i-th frame, represents the product of the two vector moduli above, Indicates the displacement distance of the shoulder coordinates from the i-th frame to the i+1-th frame in three-dimensional space, represents the difference in the elbow-wrist vector angle between the i-th frame and the i+1-th frame, is the displacement disturbance adjustment factor, is the trend adjustment factor, It indicates the number of frames with the same continuous angle change trend, and i indicates the number of the currently processed frame in the time series; The angle change fluctuation rate is obtained by calculating the standard deviation of the vector angle difference sequence between adjacent frames; The directional stability threshold is obtained by collecting multiple stable motion trajectories, calculating the standard deviation distribution of the angle change fluctuation rate, and selecting the high percentile value as the threshold; The action recognition submodule analyzes the angle combination pattern between the shoulder, elbow, and wrist vectors in the trajectory segment according to the directional stable trajectory interval, identifies the action type of the person in the training field, outputs the action type label, detects abnormal action types, and generates a posture recognition record sequence.

5. The shooting training safety system with real-time monitoring and abnormal warning according to claim 4 is characterized in that: The abnormality identification module includes: A movement trajectory extraction submodule extracts the three-dimensional coordinates of the person's shoulder in consecutive frames based on the posture recognition record, arranges the coordinate points in chronological order, identifies the spatial position change trend between consecutive coordinates, generates the movement trajectory of the shoulder in the training area, and generates a movement trajectory sequence; A path analysis submodule arranges coordinate points in chronological order according to the movement trajectory sequence, constructs a trajectory direction sequence based on the coordinate change trend, calculates the angle change between adjacent direction vectors, extracts trajectory segments with continuous angle change, extends the trajectory segments based on the end direction vectors to generate a predicted movement path sequence, and generates a path offset feature segment set; The boundary crossing prediction submodule compares the spatial position relationship between the path and the boundary based on the path offset feature segment set, combines the movement trajectory and the boundary point set, screens the path segments that overlap with the boundary, extracts the personnel number and time position corresponding to the overlapping segments, and generates a boundary crossing event sequence.

6. The shooting training safety system with real-time monitoring and abnormal warning according to claim 5 is characterized in that: The specific formula for calculating the angle change between adjacent direction vectors is: ； Calculate the angle between adjacent directions; in, Represents the direction angle between the jth frame and the j+1th frame, Represents the dot product result of the direction vector of the j-th frame and the j+1-th frame, represents the adjustment coefficient of the direction difference of the j-th frame, Represents the position offset between the jth frame and the j+1th frame, Represents the product of the direction vector modulus of the j-th frame and the j+1-th frame, represents the smoothing factor in the direction of the j-th frame, Represents the index of the current frame.

7. The shooting training safety system with real-time monitoring and abnormal warning according to claim 6 is characterized by: The signal output module obtains the corresponding number and time position in the boundary crossing event sequence, locates the abnormal path using the boundary point cloud set, outputs the warning content based on the abnormal action type recognition result, sends the warning prompt information and outputs the sound and light warning equipment control instructions to generate the abnormal behavior response record; The control instructions of the sound and light warning device are to set sound and light alarms of different frequencies according to the length of the crossing path; The real-time abnormal behavior recording unit includes a personnel number, an abnormal area identifier, and an alert response method.

8. The shooting training safety system with real-time monitoring and abnormal warning according to claim 7 is characterized in that: The signal output module includes: The event matching submodule obtains the corresponding number and time position in the cross-border event sequence, matches the personnel identification corresponding to the number, extracts the abnormal action type recognition result, organizes the behavior time sequence of the number and the action tag content, and generates the event state synchronization value; The path positioning submodule, based on the event state synchronization value, calls the spatial coordinate interval corresponding to the event time period in the boundary point cloud set, locates the spatial movement path of the person identifier within the corresponding time period, screens the coordinate point segments where the boundary crosses in the path, extracts the continuous spatial displacement trajectory, and obtains the cross-border path position sequence; The control instruction submodule extracts the corresponding warning level type information based on the out-of-bounds path position sequence and the relationship between event time and trajectory number, matches the sound and light equipment control parameters according to the warning level type, and generates an abnormal behavior response record.

9. A shooting training safety method with real-time monitoring and abnormal warning, characterized by: The method is used to implement the shooting training safety system with real-time monitoring and abnormality warning according to any one of claims 1 to 8, and the method comprises: S1: Using millimeter-wave radar and infrared positioning equipment, we acquire point cloud data from the training ground in real time, extract the 3D coordinates of multiple reference points, calculate the distances and directional angles between these points based on the point cloud spatial coordinates, classify point segments with consistent directional changes, integrate them into continuous boundary line segments, and output them as a set of training ground boundary points. S2: Obtain the boundary point set of the training ground, use point cloud data to detect the three-dimensional coordinate data of the person in the training ground, obtain the space of the shoulder, elbow, and wrist, calculate the distance and time difference between adjacent time frames, generate a velocity vector, construct a point-continuous trajectory path according to the time series, and output it as a motion trajectory feature set; S3: Obtain the motion trajectory feature set, extract the direction angle change value between each pair of adjacent frames in the shoulder, elbow, and wrist trajectory vectors, extract the trajectory segment with continuous direction change, identify the motion type and output the corresponding motion label, and output it as a posture recognition record; S4: Obtain the posture recognition record, extract the shoulder coordinate sequence, identify the movement trajectory of the person, analyze the speed change direction and the degree of continuous path deviation, combine the boundary point position, identify and predict the cross-border behavior, and output the cross-border event sequence; S5: Obtain the corresponding number and time position of the cross-border event sequence, use the boundary point cloud set to locate the abnormal path, combine the action label content in the corresponding posture recognition record, generate warning content according to the time index and abnormal path number, send warning prompt information and output control instructions for the audio-visual equipment, and generate an abnormal behavior response record.

Images