A visual and badge track joint verification method and system

By converting visual motion trajectories and ID badge positioning trajectories to a unified coordinate system in stations, the common offset of ID badge positioning trajectories is determined and corrected. The residual deviation before and after intersection is used to determine the deviation exchange, thus solving the problem of trajectory misconnection caused by visual tracking and ID badge positioning errors and improving the accuracy of violation incidents.

CN123071950APending Publication Date: 2026-11-10SICHUAN GUORUAN SCI & TECH DEV CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611565178.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-10
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

In stations, errors in visual tracking and smart badge positioning systems can lead to misalignment of staff tracks, resulting in incorrect attribution of violations. Existing technologies struggle to distinguish between regional shifts in badge positioning and individual misalignments in visual tracks.

Method used

By acquiring visual motion trajectories and employee badge positioning trajectories, and converting them to a unified coordinate system, the common trajectory offset is determined and the employee badge positioning trajectory is corrected. The deviation exchange relationship is judged by the change of residual deviation before and after the intersection, and the personnel correspondence of the visual motion trajectories is corrected.

Benefits of technology

It improves the accuracy of visual trajectory personnel correspondence and candidate violation event personnel attribution, and reduces erroneous trajectory correction and violation personnel attribution.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123071950A_ABST
    Figure CN123071950A_ABST
Patent Text Reader

Abstract

The embodiment of the application discloses a kind of visual and work card trajectory combined verification method and system, it is related to station safety monitoring technical field.The method comprises the following steps: obtaining the visual motion trajectory of multiple target personnel in the station area and corresponding work card positioning trajectory, and converting to unified coordinate system;Determine the common trajectory offset of the corresponding area, and correct each work card positioning trajectory based on the common trajectory offset;Determine the individual residual deviation between the corrected work card positioning trajectory and the corresponding visual motion trajectory;When two visual motion trajectories intersect, compare the individual residual deviation change of each visual motion trajectory relative to two corresponding work card positioning trajectories before and after intersection, determine whether deviation exchange relationship is formed;Correct the personnel corresponding relationship of visual motion trajectory, and personnel attribution verification is carried out to candidate violation event, and generate station safety risk card control information.The application improves the accuracy of visual trajectory personnel corresponding relationship and candidate violation event personnel attribution.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of railway station security monitoring technology, and in particular to a method and system for joint verification of visual and employee badge trajectories. Background Technology

[0002] Station platforms, shunting areas, and equipment operation areas are typically equipped with video surveillance equipment. By detecting and continuously tracking workers in the surveillance images, the visual motion trajectories of the workers can be formed, and candidate violations such as crossing safety boundaries or entering restricted areas can be further identified. Additionally, some stations equip workers with smart badges, which use wireless positioning to obtain the badge's location and can generate the worker's location trajectory based on the badge's binding relationship with the worker.

[0003] In practical applications, since staff members usually wear similar work clothes, safety helmets or reflective clothing, when multiple staff members approach each other, cross paths or experience short-term obstruction, target identification may be misconnected during visual tracking. That is, the visual movement trajectory of one staff member after the intersection is incorrectly continued to another staff member, resulting in subsequent candidate violations being attributed to the wrong person.

[0004] Using the location trajectory of smart work badges to assist in verifying visual motion trajectories can solve the above problems to some extent. However, stations have train bodies, metal facilities, walls, and other wireless signal obstructions, and the smart work badges themselves may also experience localized location drift. When there is a positional difference between the visual motion trajectory and the work badge location trajectory, directly using the work badge location result as the basis for judging whether the visual identity is correct may mistakenly identify the regional common offset of the work badge location as a visual tracking error, still resulting in incorrect trajectory correction and attribution of violators.

[0005] Therefore, a technical solution is needed that can distinguish between regional common offsets in work badge positioning and individual misalignments in visual trajectories, and verify the correspondence between visual personnel accordingly. Summary of the Invention

[0006] This application provides a method and system for joint verification of visual and employee ID card trajectories, which improves the accuracy of the correspondence between visual trajectories and the attribution of personnel in candidate violation events.

[0007] This application provides the following solution:

[0008] According to the first aspect, a method for joint verification of visual motion and work badge trajectories is provided, comprising: acquiring the visual motion trajectories and corresponding work badge positioning trajectories of multiple target personnel within a station area, and converting them to a unified coordinate system; determining the common trajectory offset of the corresponding area based on the positional offset between the visual motion trajectories of the multiple target personnel and the work badge positioning trajectories, and correcting each work badge positioning trajectory based on the common trajectory offset; determining the individual residual deviation between the corrected work badge positioning trajectory and the corresponding visual motion trajectory; when two visual motion trajectories intersect, comparing the changes in the individual residual deviations of each visual motion trajectory relative to the two corresponding work badge positioning trajectories before and after the intersection, and determining whether a deviation exchange relationship is formed; correcting the personnel correspondence of the visual motion trajectories based on the deviation exchange relationship, and performing personnel attribution verification on candidate violation events based on the corrected visual motion trajectories, thereby generating station safety risk control information.

[0009] According to one achievable method in this application embodiment, determining the common trajectory offset of the corresponding area based on the positional offset between the visual motion trajectories of multiple target personnel and the nameplate positioning trajectory includes: for each target personnel, obtaining multiple visual positions from the visual motion trajectory according to the sampling time, and obtaining the nameplate position corresponding to the sampling time of each visual position from the corresponding nameplate positioning trajectory; determining the nameplate positioning offset direction and nameplate positioning offset amount of the corresponding nameplate position relative to the visual position based on each visual position; comparing the nameplate positioning offsets corresponding to multiple target personnel, and determining the nameplate positioning offsets with consistent nameplate positioning offset directions and nameplate positioning offset amount differences that meet preset offset conditions as common nameplate positioning offsets; and determining the common trajectory offset of the nameplate positioning trajectory relative to the visual motion trajectory within the corresponding area based on the offset direction and offset amount of the common nameplate positioning offset.

[0010] According to one achievable method in an embodiment of this application, determining whether a deviation exchange relationship is formed includes: identifying the two intersecting visual motion trajectories as a first visual motion trajectory and a second visual motion trajectory, and identifying the badge positioning trajectories that correspond to the first and second visual motion trajectories before the intersection as a first badge positioning trajectory and a second badge positioning trajectory; determining the residual deviation of the first visual motion trajectory relative to the first and second badge positioning trajectories before the intersection, and the residual deviation of the second visual motion trajectory relative to the first and second badge positioning trajectories; determining the residual deviation of the first visual motion trajectory relative to the first and second badge positioning trajectories after the intersection, and the residual deviation of the second visual motion trajectory relative to the first and second badge positioning trajectories; comparing the changes in residual deviation before and after the intersection, and determining that the deviation exchange relationship is formed when the residual deviation of the first visual motion trajectory relative to the first badge positioning trajectory increases and the residual deviation relative to the second badge positioning trajectory decreases, and the residual deviation of the second visual motion trajectory relative to the second badge positioning trajectory increases and the residual deviation relative to the first badge positioning trajectory decreases.

[0011] According to one achievable method in an embodiment of this application, before determining the common work badge positioning offset, the method further includes: for multiple target personnel located in the same station area, determining a first relative displacement between the corresponding visual positions of any two target personnel and a second relative displacement between the corresponding work badge positions; comparing the first relative displacement and the second relative displacement corresponding to the same combination of target personnel to determine the relative displacement difference; determining the target personnel whose relative displacement differences corresponding to multiple combinations of target personnel all meet a preset consistency condition as common offset reference personnel; and determining the common work badge positioning offset based on the work badge positioning offset corresponding to the common offset reference personnel.

[0012] According to one achievable method in this application embodiment, determining the common trajectory offset of the work badge positioning trajectory relative to the visual motion trajectory within the corresponding area based on the offset direction and offset amount of the common work badge positioning offset includes: converting each of the common work badge positioning offsets into work badge positioning offset vectors according to the corresponding offset direction and offset amount; determining the common offset components of the multiple work badge positioning offset vectors in each coordinate direction of the station area coordinate system; generating a common offset vector based on the common offset components in each coordinate direction; and determining the common offset vector as the common trajectory offset of the corresponding area.

[0013] According to one achievable method in the embodiments of this application, the step of correcting the positioning trajectory of each work badge based on the common trajectory offset includes: performing reverse compensation on the trajectory position of each work badge positioning trajectory in the corresponding area according to the offset direction and offset amount corresponding to the common trajectory offset, so as to obtain the corrected work badge positioning trajectory.

[0014] According to one achievable method in an embodiment of this application, the intersection of the two visual motion trajectories includes: the distance between the two visual motion trajectories is reduced to within a preset intersection distance, and then increased again to above the preset intersection distance in subsequent continuous monitoring images.

[0015] According to one achievable method in the embodiments of this application, the step of correcting the personnel correspondence of visual motion trajectories based on the deviation exchange relationship includes: exchanging the personnel identifiers of the two visual motion trajectories that have the deviation exchange relationship after they intersect, and extending the corrected personnel identifiers to the subsequent trajectory segments of the corresponding visual motion trajectories.

[0016] According to one achievable method in the embodiments of this application, the step of verifying the personnel attribution of candidate violation events based on the modified visual motion trajectory includes: determining the time and location of the event corresponding to the candidate violation event; determining the target trajectory segment passing through the location of the event from the modified visual motion trajectory; and determining the target personnel corresponding to the candidate violation event based on the personnel identifier and work badge corresponding to the target trajectory segment.

[0017] According to the second aspect, a visual and ID badge trajectory joint verification system is provided, characterized by comprising: a trajectory acquisition module, used to acquire the visual motion trajectories of multiple target personnel and their corresponding ID badge positioning trajectories within a station area, and convert them to a unified coordinate system; a common offset determination module, used to determine the common trajectory offset of the corresponding area based on the positional offset between the visual motion trajectories of multiple target personnel and the ID badge positioning trajectories, and to correct each ID badge positioning trajectory based on the common trajectory offset; a residual deviation determination module, used to determine the individual residual deviation between the corrected ID badge positioning trajectory and the corresponding visual motion trajectory; a deviation exchange judgment module, used to compare the changes in the individual residual deviations of each visual motion trajectory relative to the two corresponding ID badge positioning trajectories before and after the intersection of two visual motion trajectories when the two visual motion trajectories intersect, and to determine whether a deviation exchange relationship has been formed; and a joint verification module, used to correct the personnel correspondence of the visual motion trajectories based on the deviation exchange relationship, and to perform personnel attribution verification on candidate violation events based on the corrected visual motion trajectories, generating station safety risk control information.

[0018] According to the specific embodiments provided in this application, the following technical effects are disclosed:

[0019] This application determines the common trajectory offset of employee badge positioning by comparing the visual motion trajectories of multiple target personnel with the badge positioning trajectory. First, it corrects the common positioning error in the badge positioning trajectory, and then uses the individual residual deviation between the corrected trajectories to identify whether deviation exchange occurs during the personnel intersection process, thereby improving the accuracy of the correspondence between visual trajectory personnel and the attribution of personnel in candidate violation events.

[0020] Of course, any product implementing this application does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A flowchart of the visual and employee badge trajectory joint verification method provided in the embodiments of this application;

[0023] Figure 2 This is a schematic diagram illustrating the common offset between the visual motion trajectory and the employee badge positioning trajectory provided in the embodiments of this application.

[0024] Figure 3 This is a schematic diagram illustrating the process of the intersection of two visual motion trajectories provided in an embodiment of this application.

[0025] Figure 4 This is a structural block diagram of the vision and employee badge trajectory joint verification system provided in the embodiments of this application. Detailed Implementation

[0026] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0027] Figure 1 This document provides a flowchart of a method for joint verification of visual data and employee badge trajectory, as illustrated in an embodiment of this application. Figure 1 As shown, the method may include the following steps:

[0028] Step 101: Obtain the visual motion trajectory and corresponding work badge positioning trajectory of multiple target personnel within the station area, and convert them to a unified coordinate system.

[0029] Step 102: Based on the positional offset between the visual motion trajectory of multiple target personnel and the work badge positioning trajectory, determine the common trajectory offset of the corresponding area, and correct the positioning trajectory of each work badge based on the common trajectory offset.

[0030] Step 103: Determine the individual residual deviation between the corrected name tag positioning trajectory and the corresponding visual motion trajectory.

[0031] Step 104: When two visual motion trajectories intersect, compare the individual residual deviation changes of each visual motion trajectory relative to the two corresponding work badge positioning trajectories before and after the intersection to determine whether a deviation exchange relationship has been formed.

[0032] Step 105: Correct the personnel correspondence of the visual motion trajectory according to the deviation exchange relationship, and perform personnel attribution verification on the candidate violation event according to the corrected visual motion trajectory to generate station safety risk control information.

[0033] As can be seen from the above process, this application determines the common trajectory offset of the nameplate positioning by comparing the visual motion trajectories of multiple target personnel with the nameplate positioning trajectory. First, the common positioning error in the nameplate positioning trajectory is corrected, and then the individual residual deviation between the corrected trajectories is used to identify whether deviation exchange occurs during the personnel intersection process, thereby improving the accuracy of the visual trajectory personnel correspondence and the attribution of candidate violation event personnel.

[0034] The following describes in detail each step of the above process and the effects that can be further produced, with reference to the embodiments.

[0035] Step 101 specifically involves: acquiring the visual motion trajectories of multiple target personnel within the station area and their corresponding work badge positioning trajectories, and converting them to a unified coordinate system.

[0036] Acquiring the visual motion trajectories and corresponding work badge positioning trajectories of multiple target personnel within a station area refers to continuously acquiring the positions of personnel entering the target monitoring area using both existing video surveillance equipment and smart work badge positioning devices. The visual motion trajectory reflects the continuous movement of the target personnel in the video surveillance image, while the work badge positioning trajectory reflects the continuous movement of the smart work badge associated with the target personnel in the actual station space. By simultaneously acquiring both types of trajectories, two sets of independent position data for the same personnel can be formed on the visual perception side and the work badge positioning side, providing a data foundation for subsequent judgment on whether personnel misconnection has occurred during visual tracking.

[0037] Specifically, the station monitoring equipment continuously acquires video images of the target area, detects personnel targets in each frame, and obtains the image positions of the target personnel at different sampling times. For the same target personnel, based on the positional continuity, motion direction continuity, and target appearance features between consecutive image frames, the personnel positions corresponding to different sampling times are sequentially connected to form the visual motion trajectory of the target personnel. The visual motion trajectory can consist of multiple visual positions arranged in chronological order, with each visual position corresponding to at least one sampling time and the spatial position of the target personnel in the monitoring image.

[0038] For the smart work badge location tracking, station staff wear smart work badges that are linked to their identity information. The smart work badges periodically generate location data via wireless positioning base stations deployed within the station, obtaining the badge's location at different sampling times. Arranging multiple work badge locations generated by the same smart work badge at consecutive sampling times in chronological order creates a corresponding work badge location track. Since the smart work badge is already linked to the staff member, the work badge location track can serve as an alternative source of location changes for that staff member, independent of visual monitoring results.

[0039] The corresponding smart badge location trajectory refers to obtaining the location data of the smart badge corresponding to the current target person based on the existing binding relationship between the target person and the smart badge. For example, before a staff member enters the target station area, the station personnel management system can pre-record the binding relationship between the staff member's number and the badge number. After obtaining the visual motion trajectory of a target person, the corresponding smart badge location trajectory can be determined based on the person's existing identity information, the initial binding relationship when entering the area, or the historical stable trajectory binding relationship.

[0040] Since visual motion trajectories are typically generated in the coordinate system of the surveillance image, while employee badge positioning trajectories are typically generated in the planar coordinate system corresponding to the station positioning system, the two types of trajectories cannot be directly compared in their original state. To give visual positions and employee badge positions a unified positional meaning, the two types of trajectories need to be transformed to a unified coordinate system.

[0041] Figure 2 This diagram illustrates the common offset between the visual motion trajectory and the employee badge positioning trajectory provided in an embodiment of this application. In one implementation, the station planar area coordinate system can be used as a unified coordinate system. Multiple fixed and easily identifiable reference positions within the station area are pre-selected. The image coordinates of each reference position in the monitoring image and its actual coordinates in the station planar area coordinate system are determined. A coordinate transformation relationship is established based on the correspondence between the two sets of coordinates. Using this coordinate transformation relationship, each visual position in the visual motion trajectory is converted into a position in the station planar area coordinate system.

[0042] For the work badge positioning trajectory, if the work badge positioning system has adopted the station plane area coordinate system, the corresponding work badge position can be used directly; if the work badge positioning system adopts an independent positioning coordinate system, the coordinate transformation of each work badge position is performed according to the known position correspondence between the work badge positioning system and the station plane area coordinate system, so that the visual position and the work badge position are both represented as the actual spatial position in the same station area.

[0043] Furthermore, to ensure the comparability of subsequent position offset calculations, a temporal mapping can be performed between the visual motion trajectory and the employee badge positioning trajectory. Since the sampling frequencies of the video surveillance equipment and the employee badge positioning equipment may differ, the location with the closest sampling time can be selected for mapping. Alternatively, position interpolation can be performed on trajectories with lower sampling frequencies, ensuring that the visual position and employee badge position form a position pair at a unified sampling time. After temporal mapping and spatial coordinate unification, each target person can obtain a corresponding visual position sequence and employee badge position sequence.

[0044] Through the above processing, visual motion trajectories and ID badge positioning trajectories that originally had different sources, coordinate definitions, and sampling frequencies can be converted into trajectory data that can be directly compared spatially. This provides a unified data foundation for subsequently determining the common trajectory offset of the ID badge positioning trajectory relative to the visual motion trajectory, calculating individual residual deviations, and identifying deviation exchange relationships during the intersection process.

[0045] Step 102 specifically involves: determining the common trajectory offset of the corresponding area based on the positional offset between the visual motion trajectory of multiple target personnel and the work badge positioning trajectory, and correcting the positioning trajectory of each work badge based on the common trajectory offset.

[0046] Determining the common trajectory offset for a corresponding area based on the positional offset between the visual motion trajectories and the ID badge positioning trajectories of multiple target personnel involves identifying potential regional overall drift in the smart ID badge positioning data by utilizing the common positional offset characteristics exhibited between the two types of trajectories of multiple target personnel within the same area, after the visual motion trajectories and ID badge positioning trajectories have been transformed to a unified station area coordinate system. The purpose of this process is not to directly correct the visual position using the ID badge position of a single target personnel, but rather to first determine whether the ID badge positioning results of multiple target personnel simultaneously exhibit approximately directional and degree of offset relative to the visual position, and then extract the common trajectory offset that characterizes the overall offset state of the ID badge positioning system in that area.

[0047] Specifically, for each target person, multiple visual positions are selected from their visual motion trajectory according to the sampling time. Simultaneously, the nameplate position corresponding to the sampling time of each visual position is extracted from the nameplate positioning trajectory corresponding to that target person. Thus, each visual position has a temporally corresponding nameplate position. The correspondence between sampling times can be achieved by directly corresponding to the same sampling time, or, when the sampling frequencies of the two types of data are different, a correspondence can be established based on temporal proximity or interpolation results, thereby reducing positional differences caused by different sampling times.

[0048] For each corresponding visual position and badge position, the visual position is used as the positional reference to determine the badge positioning offset relative to the visual position. The badge positioning offset includes at least the badge positioning offset direction and the badge positioning offset amount. The badge positioning offset direction represents the spatial direction from the visual position to the badge position, and the badge positioning offset amount represents the spatial distance between the visual position and the badge position. Since the visual position is used as the reference, the badge positioning offset directly describes the positional shift of the smart badge positioning result relative to the visual detection result, rather than a simple, undirected comparison of the positional differences between the two types of trajectories.

[0049] For example, if the visual position of the same target person at a certain sampling time is located at the first position in the station area coordinate system, and the corresponding name tag position is located at a certain distance to the right and front of that visual position, then the name tag positioning offset direction at that sampling time can be determined to be from the visual position to the right and front, and the name tag positioning offset amount is the actual distance between the two positions. Following the same method, multiple name tag positioning offsets corresponding to the same target person at multiple sampling times can be continuously obtained, and further, the name tag positioning offsets corresponding to multiple target persons within the same area can be obtained separately.

[0050] In a real-world station environment, the wireless positioning signal of smart work badges may be affected by train bodies, metal structures, platform structures, or other environmental factors, causing multiple smart work badges within the same area to drift simultaneously in approximately the same direction. If the judgment is based solely on the distance between the visual position of a single target person and the position of their work badge, it is difficult to distinguish whether this difference is due to regional drift generated by the work badge positioning system, visual tracking errors, individual work badge anomalies, or other reasons. Therefore, this embodiment further filters personnel used to determine common trajectory deviations by analyzing the positional relationships between multiple target personnel.

[0051] Specifically, for multiple target personnel located within the same station area, any two target personnel are selected to form a target personnel combination. A first relative displacement between the two visual positions corresponding to this target personnel combination is determined, and a second relative displacement between the two corresponding name tag positions is also determined. The first relative displacement represents the relative spatial relationship between the two target personnel in the visual detection results, and the second relative displacement represents the relative spatial relationship between the same two target personnel in the name tag positioning results.

[0052] The relative displacement reflects not only the distance between two target personnel, but also the direction of one target personnel relative to the other. Therefore, when multiple smart badges experience overall positional drift, but the relative spatial structure between different badges remains essentially unchanged, the corresponding first and second relative displacements will still maintain a high degree of consistency.

[0053] For example, three individuals move forward sequentially at a certain distance from each other in their visual positions, while the positions of the three name tags as a whole shift in the same direction relative to their corresponding visual positions. If all three name tags experience approximately a uniform positioning drift, then the relative distance and direction between any two individuals in the name tag positioning data should still be essentially consistent with their relative distance and direction in the visual data. Therefore, this relationship of maintaining relative spatial structure can be used as an important criterion for determining whether the name tag shift is a common shift.

[0054] Furthermore, the first and second relative displacements corresponding to the same group of target personnel are compared to obtain the relative displacement difference. The relative displacement difference can be determined based on the relative distance difference, the relative direction difference, or a combination of both. When the relative displacement differences corresponding to multiple groups of target personnel all meet the preset consistency condition, it indicates that the relative spatial structure of the multiple target personnel is consistent in the visual data and the name tag positioning data. Although the name tag position may shift relative to the visual position, this shift does not significantly change the mutual positional relationship between the multiple target personnel.

[0055] In this case, target personnel who meet the above conditions are identified as common offset reference personnel. Common offset reference personnel provide a reliable data source for determining regional name tag positioning offsets. By first filtering common offset reference personnel using the relative spatial structure of multiple individuals, the adverse effects on the common trajectory offset calculation results caused by visual tracking errors, name tag positioning anomalies, or trajectory correspondence errors occurring in individual target personnel can be reduced.

[0056] For common offset reference personnel, their corresponding work badge positioning offsets are further compared. If the work badge positioning offset directions of multiple common offset reference personnel are consistent, and the differences between the work badge positioning offsets meet the preset offset conditions, then these work badge positioning offsets are determined as common work badge positioning offsets.

[0057] Consistent job badge positioning offset directions do not require all offset directions to be exactly the same. The allowable range of directional differences can be set based on the actual positioning accuracy. When the differences between multiple offset directions are within the allowable range, they can be considered to have consistent offset directions. Similarly, job badge positioning offset differences meeting preset offset conditions mean that the differences between multiple job badge positioning offsets do not exceed the allowable offset difference range. By simultaneously limiting both offset direction and offset amount, it is possible to avoid classifying multiple job badges as having common positioning drift simply because they all offset in roughly the same direction.

[0058] The shared work badge positioning offset reflects the common offset state exhibited simultaneously by multiple reference personnel. It is primarily used to further extract the common trajectory offset of the work badge positioning system within the current station area. Compared to the positioning offset of a single person's work badge, the shared work badge positioning offset can reduce the impact of individual positioning noise and localized personnel movement, making it more suitable for representing regional overall positioning errors.

[0059] After determining the positioning offset of the shared work badge, the positioning offset of each shared work badge is converted into a work badge positioning offset vector according to its offset direction and offset amount. In vector form, the position offset, which was originally described by both direction and distance, can be converted into coordinate direction components in the station area coordinate system, thus facilitating the unified calculation of the positioning offsets of multiple shared work badges.

[0060] For example, when using a two-dimensional station area coordinate system, the positioning offset vector of each work badge can be decomposed into an offset component in the first coordinate direction and an offset component in the second coordinate direction. If a three-dimensional station area coordinate system is used, an offset component in the height direction can also be included. This embodiment does not limit the specific dimensions of the unified coordinate system and can be set according to the actual deployment method of the station positioning system.

[0061] The offset components of multiple work badge positioning offset vectors in each coordinate direction of the station area coordinate system are statistically analyzed, and the common offset component in the corresponding coordinate direction is determined based on the multiple offset components. The common offset component can be determined by the average value of the corresponding components of multiple common work badge positioning offsets, or by using the median or other methods that can characterize the central tendency of multiple offset components. The purpose is to extract the regional offset portion shared by multiple common offset reference personnel, while reducing the impact of individual work badge positioning errors on the results.

[0062] After obtaining the common offset components in each coordinate direction, these components are combined to form a common offset vector, which is then defined as the common trajectory offset of the smart badge positioning trajectory relative to the visual motion trajectory within the corresponding station area. This common trajectory offset can be understood as the overall spatial drift of the smart badge positioning result relative to the visual detection position within the current station area.

[0063] For example, if the ID badge positions of multiple common offset reference personnel are all mainly offset laterally to the right of the station relative to their visual positions, accompanied by a small longitudinal offset, then the lateral and longitudinal common offset components can be determined separately, and a corresponding common offset vector can be formed accordingly. This common offset vector can characterize the overall positional drift jointly generated by the ID badge positioning systems within the current area.

[0064] After determining the common trajectory offset, reverse compensation is performed on the positioning trajectories of each work badge within the current area based on the offset direction and amount corresponding to the common trajectory offset. This reverse compensation refers to moving the position of each work badge's positioning trajectory by a distance corresponding to the common trajectory offset in the opposite direction, thereby offsetting the regional common positioning drift.

[0065] When using a common offset vector to represent the common trajectory offset, the compensated employee badge position can be obtained by directly subtracting the common offset vector from the position coordinates corresponding to each badge trajectory position. By sequentially applying the same processing to multiple trajectory positions in the employee badge positioning trajectory, the corrected employee badge positioning trajectory can be formed.

[0066] For example, if it is determined that the public trajectory offset within the current station area manifests as an overall rightward shift in the ID card positioning result, then when correcting the ID card positioning trajectory, the position of each ID card trajectory will be compensated to the left by the corresponding distance. If the public trajectory offset includes both lateral and longitudinal offsets, then reverse compensation will be performed in both coordinate directions respectively, so that the corrected ID card positioning trajectory can eliminate the overall position drift caused by the regional positioning environment as much as possible.

[0067] It should be noted that the correction target for the common trajectory offset is the smart badge positioning trajectory, not the visual motion trajectory. This is because, in this embodiment, the badge positioning offset is determined based on the visual position, and the common trajectory offset is used to describe the positioning drift caused by multiple smart badges relative to the visual motion trajectory. Therefore, by performing reverse compensation on the badge positioning trajectory, the corrected badge positioning trajectory can be made to be in a more consistent spatial reference state with the visual motion trajectory.

[0068] Different locations within a station are affected to varying degrees by the train body, metal equipment, walls, and the distribution of wireless positioning base stations. Therefore, the common offset generated by the badge positioning may not maintain the same direction and amount throughout the entire station area. If a fixed common offset vector is used to uniformly compensate for the badge positioning trajectory throughout the entire area, significant correction errors may still occur near locations where the badge positioning environment changes.

[0069] Therefore, in one implementation, the common trajectory offset corresponding to different positions is continuously determined along the movement direction of multiple target personnel, and the common trajectory offset direction and common offset amount corresponding to adjacent positions are compared. When the change in the common trajectory offset direction reaches a preset direction change condition, or the change in the common offset amount reaches a preset offset change condition, the corresponding position is determined as the common offset turning point.

[0070] Based on multiple common offset turning points, the station area traversed by the target personnel is divided into multiple offset correction sections. The common trajectory offset within the same offset correction section remains relatively stable, while adjacent offset correction sections can use different common trajectory offsets.

[0071] For a work badge trajectory position located within a certain offset correction segment, reverse compensation is performed using the common trajectory offset corresponding to that offset correction segment. For a work badge trajectory position close to the boundary between two offset correction segments, the common trajectory offset corresponding to the two adjacent offset correction segments can be transitionally calculated based on the distance from the work badge trajectory position to the two adjacent common offset turning points, thereby obtaining the target common trajectory offset corresponding to the current position.

[0072] For example, when staff enter the platform area from the station hall, the work badge positioning may gradually change from a rightward offset to a rightward and forward offset. In this case, instead of treating the entire movement process as a fixed offset area, an offset turning point is formed at the location where the common offset changes significantly, and the work badge trajectory is corrected by using the common trajectory offset corresponding to the two segments before and after the change.

[0073] Through the above processing, the public trajectory offset can be expanded from a single fixed vector into a segmented offset correction result that varies along the personnel movement path, making it more adaptable to the changes in the wireless positioning environment corresponding to different spatial locations of the station.

[0074] After the above processing, the corrected ID badge positioning trajectory has largely eliminated the regional positioning offset shared by multiple smart ID badges within the same area. At this point, the remaining positional differences between the corrected ID badge positioning trajectory and the corresponding visual motion trajectory are better able to reflect the trajectory differences at the individual target personnel level, providing a more reliable data foundation for subsequently determining individual residual deviations and judging whether a deviation exchange relationship occurs after the intersection of two visual motion trajectories.

[0075] Within the actual station area, the staff members in the same area are not always the same; some staff members may leave the current area, while new staff members may enter. If the common trajectory offset is recalculated independently based on the current staff members at each moment, changes in the reference staff set can easily cause abrupt changes in the common trajectory offset between adjacent time periods, resulting in discontinuities in the work badge positioning trajectory correction results.

[0076] Therefore, in one implementation, the continuous monitoring process is divided into multiple adjacent offset update periods, and a common offset reference person is determined for each offset update period. When the common offset reference person corresponding to two adjacent offset update periods changes, an overlapping reference person who exists in both periods is determined from the common offset reference persons of the two periods.

[0077] For each overlapping reference person, the work badge positioning offset for each person is determined in the previous and subsequent offset update periods, and the amount of offset change between the two periods is also determined. When the offset change directions for multiple overlapping reference persons are consistent and the differences between the offset changes meet the preset continuation conditions, the common trajectory offsets corresponding to the two offset update periods are considered to have a continuous change relationship.

[0078] In this case, instead of directly using the common trajectory offset calculated independently in the next offset update period, the common trajectory offset corresponding to the previous offset update period is used as the basis. The common offset update amount is determined based on the offset change amount corresponding to multiple overlapping reference personnel, and the common trajectory offset corresponding to the next offset update period is obtained using the common offset update amount.

[0079] For example, in the previous period, staff members A, B, and C were used as common offset references. In the next period, staff member A left and staff member D entered, so staff members B, C, and D became common offset references. In this case, staff members B and C, who existed in both periods, can be used to determine the change in the common offset between the two periods, and then this change can be passed on to the next period, instead of re-establishing a completely independent offset benchmark because the set of reference staff changes.

[0080] Through the above processing, the temporal continuity of the common trajectory offset can be maintained in the dynamic environment of the station where the reference personnel are constantly changing, and the jump in the work badge trajectory correction caused by the entry or exit of reference personnel can be reduced.

[0081] First, the common offset reference personnel are screened using the relative spatial structure of multiple people. Then, the common trajectory offset is extracted based on the common offset of the name tag positioning offset exhibited by multiple common offset reference personnel. Reverse compensation is then performed on the name tag positioning trajectory, which can separate the regional common error from the trajectory anomaly of a single person in the intelligent name tag positioning system.

[0082] Step 103 specifically involves determining the individual residual deviation between the corrected name tag positioning trajectory and the corresponding visual motion trajectory.

[0083] Determining the individual residual deviation between the corrected ID badge positioning trajectory and the corresponding visual motion trajectory refers to further identifying the remaining positional differences between the visual motion trajectory and the ID badge positioning trajectory for a single target person after reverse compensation using common trajectory offset. Since the aforementioned common trajectory offset has already been used to eliminate the regional positioning drift exhibited by multiple ID badges within the same station area, the remaining positional differences at this point can more effectively reflect the degree of local inconsistency between the two types of trajectories for a single target person.

[0084] Specifically, for any target person, multiple consecutive visual positions are selected from their visual motion trajectory, and the corrected badge position corresponding to the sampling time of each visual position is obtained from the corrected badge positioning trajectory. For each set of sampling times, the spatial position difference between the visual position and the corrected badge position is determined, thereby obtaining the single-point residual deviation corresponding to that sampling time.

[0085] Single-point residual bias can be determined based on the spatial distance between the visual position and the corrected work tag position. For example, in a two-dimensional station area coordinate system, the positional differences between the visual position and the corrected work tag position in two coordinate directions can be determined, and the corresponding spatial distance can be determined based on these two positional differences. The smaller the spatial distance, the closer the corrected work tag position is to the visual position, and the higher the consistency between the two types of trajectories at that sampling time; the larger the spatial distance, the more obvious the positional separation between the two types of trajectories.

[0086] In one implementation, to avoid the influence of instantaneous positioning fluctuations or visual detection errors on a single sampling position, single-point residual deviations corresponding to multiple sampling times can be continuously acquired, and the individual residual deviation of the corresponding trajectory segment can be determined based on the multiple single-point residual deviations. The individual residual deviation can be determined by the average value, median value, or other statistical measures used to represent the overall positional differences of continuous trajectory segments from the multiple single-point residual deviations.

[0087] For example, by obtaining the single-point residual deviations at five sampling times within a preset continuous time period, the individual residual deviation for that continuous time period can be determined based on the average of the five single-point residual deviations. By using multiple consecutive sampling locations for calculation, the impact of visual detection position jitter or instantaneous badge positioning error at a certain moment on trajectory correspondence judgment can be reduced.

[0088] Furthermore, the trajectory segment used to determine the individual residual bias can be selected based on the target person's movement status. For example, when the target person is moving continuously, multiple consecutive sampling locations can be selected for calculation; when the target person is in a brief stop state, multiple location data during the stop period can be selected to determine the individual residual bias. The key is to reflect the actual spatial difference between the corresponding visual motion trajectory and the corrected name tag positioning trajectory.

[0089] It should be noted that individual residual bias is determined after deducting the common trajectory offset. Suppose multiple smart badges within a certain area all deviate significantly in the same direction relative to their visual positions. If we directly compare the distance between the original badge position and the visual position, this distance includes both regional common positioning error and local differences corresponding to individual individuals. After common trajectory offset correction, the regional common offset is weakened or eliminated. The remaining individual residual bias is then primarily used to describe the inconsistency between the visual trajectory and the badge trajectory at the individual individual level.

[0090] When the target personnel move normally and the visual tracking relationship is correct, the visual motion trajectory corresponding to the same target personnel and the corrected name tag positioning trajectory usually have similar positional change trends, and their individual residual deviations should remain within a relatively small and stable range. When the visual tracking system incorrectly extends the visual personnel identification to another person due to two workers approaching, obstructing, or intersecting, the positional relationship between the original visual motion trajectory and the original corresponding name tag positioning trajectory will change, resulting in a significant increase in their individual residual deviations.

[0091] To facilitate subsequent determination of the deviation exchange relationship, the individual residual deviation of each visual motion trajectory relative to the two corrected name tag positioning trajectories can be calculated for the two intersecting visual motion trajectories. That is, for the first visual motion trajectory, its residual deviation relative to the first name tag positioning trajectory and the second name tag positioning trajectory can be determined separately; similarly, for the second visual motion trajectory, its residual deviation relative to the first name tag positioning trajectory and the second name tag positioning trajectory can be determined separately.

[0092] Through the above cross-calculation, a complete residual deviation correspondence can be established between the two visual motion trajectories and the two work badge positioning trajectories. If the residual deviation between the first visual motion trajectory and the first work badge positioning trajectory is small before the personnel converge, but the residual deviation between the first visual motion trajectory and the second work badge positioning trajectory is large, it indicates that the first visual motion trajectory mainly corresponds to the first work badge positioning trajectory before the convergence. The same method can be used to determine the main correspondence before the convergence of the second visual motion trajectory.

[0093] After the personnel converge, the residual deviation of the individuals mentioned above is re-determined. If the residual deviation of the first visual motion trajectory relative to the original first work badge positioning trajectory increases, while the residual deviation relative to the second work badge positioning trajectory decreases, and the second visual motion trajectory shows a corresponding reverse change, then it can be further used to determine whether the two visual motion trajectories form a deviation exchange relationship.

[0094] Therefore, individual residual bias is not simply used to represent the absolute distance between two trajectories, but rather as a measure of personnel-level trajectory differences after eliminating regional work tag co-location drift. By comparing the individual residual bias of the same visual motion trajectory relative to different work tag positioning trajectories, and by comparing the changes in individual residual bias before and after personnel intersection, a more reliable data basis can be provided for identifying whether visual personnel identification has been misaligned.

[0095] Step 104 specifically involves: when two visual motion trajectories intersect, comparing the individual residual deviation changes of each visual motion trajectory relative to the two corresponding work badge positioning trajectories before and after the intersection to determine whether a deviation exchange relationship has been formed.

[0096] When two visual motion trajectories intersect, the individual residual deviation changes of each visual motion trajectory relative to the two corresponding name tag positioning trajectories before and after the intersection are compared to determine whether a deviation exchange relationship has been formed. This refers to establishing a residual deviation correspondence between the visual motion trajectory and the name tag positioning trajectory before and after the personnel intersect, in scenarios where personnel approach and cross paths are prone to causing visual target tracking misalignment. Based on whether the correspondence has been mutually transferred, it is determined whether personnel identification has been exchanged during the visual tracking process.

[0097] Figure 3 This diagram illustrates the intersection process of two visual motion trajectories provided in this embodiment. Specifically, during station monitoring, the visual motion trajectories of each target person are continuously acquired, and the spatial distance between different visual motion trajectories is calculated. When the distance between two visual motion trajectories gradually decreases to within a preset intersection distance, it indicates that the corresponding two target persons have entered a state of visual tracking confusion. At this time, the positions of the two target persons in the monitoring image are relatively close, and there may be situations where human body outlines overlap, mutual occlusion, or difficulty in distinguishing appearance features.

[0098] In subsequent continuous monitoring images, the two visual motion trajectories are continued to be tracked. When the distance between the two visual motion trajectories increases again to above the preset intersection distance, it indicates that the corresponding target person has completed the approach process and separated again. The process from when the two visual motion trajectories first enter within the preset intersection distance to when they increase again to above the preset intersection distance can be regarded as a visual motion trajectory intersection process.

[0099] The preset intersection distance can be pre-set based on the camera installation location, monitoring angle, typical size of the target person in the image, and spatial resolution capability of the visual tracking algorithm. It is primarily used to filter out inter-personnel intersection scenarios where visual identity mismatch may occur, eliminating the need for subsequent deviation exchange judgments on distant target persons, thereby reducing unnecessary data processing.

[0100] Once the intersection of two visual motion trajectories is confirmed, they are designated as the first visual motion trajectory and the second visual motion trajectory, respectively. To clarify the personnel correspondence between the two visual motion trajectories before their intersection, the identification badge positioning trajectory, which had already maintained a stable correspondence with the personnel before the intersection occurred, is further obtained.

[0101] Specifically, the badge positioning trajectory that maintains a correspondence with the first visual motion trajectory before the intersection is determined as the first badge positioning trajectory, and the badge positioning trajectory that maintains a correspondence with the second visual motion trajectory before the intersection is determined as the second badge positioning trajectory. Maintaining a correspondence means that, within multiple consecutive sampling moments before the intersection, the individual residual deviation between the corresponding visual motion trajectory and the corresponding badge positioning trajectory remains within a small range, and the positional change trends of both are continuous.

[0102] The above method establishes an initial correspondence before the intersection, whereby the first visual motion trajectory corresponds to the first work badge positioning trajectory, and the second visual motion trajectory corresponds to the second work badge positioning trajectory. This initial correspondence serves as the benchmark for subsequent judgments regarding whether a deviation exchange has occurred.

[0103] Before the intersection occurs, the residual deviations of the first visual motion trajectory relative to the first and second work badge positioning trajectories are calculated respectively, and the residual deviations of the second visual motion trajectory relative to the first and second work badge positioning trajectories are calculated respectively.

[0104] In other words, before the intersection, the residual deviation between the visual motion trajectory and the original corresponding work tag positioning trajectory is not only calculated, but a cross-comparison is performed between the two visual motion trajectories and the two work tag positioning trajectories. This yields four sets of residual deviations, which describe the spatial proximity relationship between the two visual motion trajectories and the two work tag positioning trajectories, respectively.

[0105] Under normal circumstances, before they intersect, the residual deviation of the first visual motion trajectory relative to the first work tag positioning trajectory should be less than its residual deviation relative to the second work tag positioning trajectory, and the residual deviation of the second visual motion trajectory relative to the second work tag positioning trajectory should be less than its residual deviation relative to the first work tag positioning trajectory. This indicates that the two visual motion trajectories maintain a relatively stable spatial correspondence with their original corresponding work tag positioning trajectories before they intersect.

[0106] After the two target personnel complete their intersection and then separate again, the same method is used to re-determine the four sets of residual deviations. Specifically, the residual deviations of the first visual motion trajectory relative to the first and second work badge positioning trajectories after the intersection, as well as the residual deviations of the second visual motion trajectory relative to the first and second work badge positioning trajectories, are determined respectively.

[0107] By comparing the residual deviation changes of the same visual motion trajectory before and after the intersection with the two work badge positioning trajectories, it can be determined whether the visual motion trajectory still corresponds to the original target person.

[0108] For example, if the residual deviation of the first visual motion trajectory relative to the first work badge positioning trajectory is small before the intersection, but the residual deviation relative to the second work badge positioning trajectory is large, but the residual deviation relative to the first work badge positioning trajectory increases after the intersection, while the residual deviation relative to the second work badge positioning trajectory decreases, it indicates that the first visual motion trajectory gradually deviates from the original corresponding first work badge positioning trajectory after the intersection and turns to approach the second work badge positioning trajectory.

[0109] Similarly, if the residual deviation of the second visual motion trajectory relative to the second work badge positioning trajectory increases after the intersection, while the residual deviation relative to the first work badge positioning trajectory decreases, it indicates that the second visual motion trajectory also shows a trend of shifting from the original second work badge positioning trajectory to the first work badge positioning trajectory.

[0110] When the above two sets of changes occur simultaneously, a deviation exchange relationship can be established. This deviation exchange relationship indicates the correspondence between the two visual motion trajectories and the two employee badge positioning trajectories, which are paired and exchanged before and after the personnel intersect. Specifically, the first visual motion trajectory changes from primarily corresponding to the first employee badge positioning trajectory to primarily approaching the second employee badge positioning trajectory, while the second visual motion trajectory changes from primarily corresponding to the second employee badge positioning trajectory to primarily approaching the first employee badge positioning trajectory.

[0111] This deviation exchange relationship differs from the increase in deviation of a single visual motion trajectory. The increase in residual deviation of a single trajectory may be caused by instantaneous visual detection errors, jitter in the positioning of a single name tag, or short-term movement of the target person. However, the simultaneous occurrence of corresponding deviation transfers in two visual motion trajectories is more indicative of a misconnection of personnel during the convergence process.

[0112] For example, before the intersection, the residual deviation of the first visual motion trajectory relative to the first work badge positioning trajectory is relatively small, while the residual deviation relative to the second work badge positioning trajectory is relatively large; the second visual motion trajectory has the opposite deviation relationship. After the intersection, if the residual deviation of the first visual motion trajectory relative to the second work badge positioning trajectory decreases to a small range, and the residual deviation of the second visual motion trajectory relative to the first work badge positioning trajectory also decreases to a small range, then it can be considered that the two visual motion trajectories have undergone an identity exchange during the intersection process.

[0113] In one implementation, to reduce the impact of fluctuations in individual sampling times on the judgment result, multiple consecutive sampling positions before and after the intersection can be selected to determine the residual bias. For example, multiple consecutive sampling times can be selected as the pre-intersection comparison segment before the two visual motion trajectories enter a preset intersection distance, and multiple consecutive sampling times can be selected as the post-intersection comparison segment after the two visual motion trajectories increase to above the preset intersection distance again. The corresponding residual bias is determined based on the average or median value of the residual biases at multiple single points within each comparison segment.

[0114] In another embodiment, minimum changes can be set for the increase and decrease of residual deviations. The corresponding changes are only used to determine the deviation exchange relationship when the increase in the original corresponding residual deviation after the intersection reaches a preset change, and the decrease in the corresponding residual deviation after the intersection also reaches a preset change. This method avoids unstable judgments due to small changes in the distance between the two trajectories.

[0115] It should be noted that the deviation exchange relationship determination is performed after the work badge positioning trajectory has been corrected based on the common trajectory offset. Therefore, the residual deviations used for comparison have minimized the common positioning drift generated by the intelligent work badge positioning system in the current station area. This avoids the distance between the visual motion trajectory and the original work badge trajectory changing simultaneously due to multiple work badges shifting in the same direction, thereby improving the reliability of the deviation exchange relationship determination.

[0116] Through the above processing, this embodiment does not simply select the nearest work badge positioning trajectory as the new corresponding trajectory after personnel convergence. Instead, it simultaneously examines four sets of residual deviations before and after convergence and their paired changes. Only when the two visual motion trajectories show a mutual transfer phenomenon of increased original corresponding deviation and decreased cross-corresponding deviation, is a deviation exchange relationship determined, thus providing a basis for subsequent correction of the personnel correspondence of visual motion trajectories.

[0117] Step 105 specifically involves: correcting the personnel correspondence of the visual motion trajectory based on the aforementioned deviation exchange relationship, and verifying the personnel attribution of candidate violation events based on the corrected visual motion trajectory, thereby generating station safety risk control information.

[0118] Correcting the personnel correspondence of visual motion trajectories based on deviation exchange relationships refers to the process of redistributing personnel identifications on visual motion trajectories after they intersect, based on the correspondence between the visual motion trajectories and the work badge positioning trajectories established before the intersection. This ensures that the visual motion trajectories are reassigned to the target personnel actually carrying the corresponding smart work badges.

[0119] Specifically, before the two visual motion trajectories intersect, the first visual motion trajectory maintains a stable correspondence with the first employee badge positioning trajectory, and the second visual motion trajectory maintains a stable correspondence with the second employee badge positioning trajectory. If, after the intersection, the individual residual deviation of the first visual motion trajectory relative to the second employee badge positioning trajectory decreases, and simultaneously the individual residual deviation of the second visual motion trajectory relative to the first employee badge positioning trajectory decreases, forming the aforementioned deviation exchange relationship, it indicates that the visual tracking system may have misconnected the subsequent visual trajectories of the two target personnel during the intersection process.

[0120] In this scenario, the personnel identifiers at the intersection of the two visual motion trajectories are swapped. That is, the personnel identifiers that originally extended to the trajectory segment after the first visual motion trajectory intersects are changed to the personnel identifiers corresponding to the second target personnel, and the personnel identifiers that originally extended to the trajectory segment after the second visual motion trajectory intersects are changed to the personnel identifiers corresponding to the first target personnel, thereby restoring the correct correspondence between the visual motion trajectories and the actual personnel.

[0121] Personnel identification can be a trajectory number used within the visual tracking system to distinguish different target personnel, or it can be a staff identity identifier obtained through a smart badge binding relationship. Regardless of the specific form of identification used, its purpose is to ensure that the converged visual trajectory data can continue to be attributed to the actual corresponding target personnel.

[0122] After the personnel identification exchange is completed in the trajectory segment following the intersection, the corrected personnel identification is carried over to the subsequent trajectory segments of the corresponding visual motion trajectory. In other words, as long as no new personnel intersections or deviation exchanges are detected, subsequent continuous visual trajectories will inherit the current corrected personnel identification to avoid the erroneous identification generated by the visual tracking system in the previous intersection from continuing to affect subsequent trajectory data.

[0123] For example, before the convergence, visual trajectory A corresponds to worker A, and visual trajectory B corresponds to worker B. If, after the convergence, the visual tracking system incorrectly labels worker A's subsequent actual trajectory as B and worker B's subsequent actual trajectory as A, then after detecting the deviation exchange relationship, the labels of the two trajectories after the convergence are swapped, so that worker A is reassigned to his actual subsequent visual trajectory, and worker B is reassigned to another actual subsequent visual trajectory.

[0124] Through the above processing, the visual motion trajectory can not only maintain continuity in spatial location, but also restore continuity in the correspondence of personnel identities, avoiding the continuous propagation of visual identity misconnection caused by a single intersection to the subsequent monitoring process.

[0125] The aforementioned personnel attribution verification based on the corrected visual motion trajectory refers to the process where, after the visual monitoring system has identified a candidate violation, instead of directly using the uncorrected visual personnel identifier to determine the responsible personnel, the system re-determines the actual target personnel corresponding to the candidate violation based on the visual motion trajectory corrected by deviation exchange.

[0126] Specifically, the time and location of the candidate violation event are first determined. The time of occurrence can be determined based on the time when the visual monitoring system first identifies the target person who meets the candidate violation criteria, or it can be determined based on a representative moment during the duration of the candidate violation. The location of occurrence can be determined based on the visual position of the target person corresponding to the candidate violation event in a unified station area coordinate system.

[0127] Candidate violations may include staff entering restricted areas, crossing preset safety boundaries, entering hazardous work areas, or other abnormal behaviors detected by the station's visual safety monitoring system. For this invention, the specific identification method for candidate violations can utilize existing station visual safety monitoring methods; the focus of this invention is on verifying the attribution of personnel corresponding to candidate violations.

[0128] After determining the time and location of the event, the visual trajectory segment that passes through the event location at the corresponding time is found from the corrected visual motion trajectory, and this visual trajectory segment is determined as the target trajectory segment. To improve matching stability, an allowable position range can be set around the event location, and the target trajectory segment can be determined by combining the continuous visual positions before and after the event time.

[0129] When multiple visual trajectories are close to the location of the incident, the target trajectory segment that best matches the candidate violation incident can be determined based on the location distance, direction of movement, and trajectory continuity at the time of the incident, so as to avoid interference from nearby staff in judging the attribution of the incident.

[0130] After determining the target trajectory segment, the corrected personnel identifier for that trajectory segment is obtained, and the target personnel are identified based on the smart badge corresponding to that identifier. Since the smart badge is pre-linked with station staff, the staff number, job number, or other information that can distinguish the staff member's identity can be further obtained through the badge corresponding to the target trajectory segment.

[0131] By simultaneously utilizing the corrected visual trajectory personnel identification and the employee badge binding relationship, the incorrect attribution of candidate violations due to relying solely on visual tracking identification can be avoided. Especially when a candidate violation occurs after two employees have met, even if the original visual tracking system has already exchanged personnel identifications, the aforementioned deviation exchange relationship can be used to restore the correct personnel correspondence and then determine the actual target personnel of the candidate violation.

[0132] For example, after two staff members meet in the platform area, one of them enters the risk area, and the visual monitoring system generates a candidate violation event. If the original visual tracking result shows that personnel identification has already been exchanged when the two meet, directly reading the original visual trajectory identification may incorrectly attribute the candidate violation event to the other staff member. This embodiment first identifies the deviation exchange relationship through the work badge positioning trajectory and individual residual deviation, and then corrects the visual personnel identification after the meeting. Therefore, it can re-attribute the target trajectory segment corresponding to the location of the event to the staff member who actually entered the risk area.

[0133] After identifying the target personnel corresponding to the candidate violation incident, station safety risk control information is generated based on the target personnel, the time of the incident, the location of the incident, and relevant information about the candidate violation incident. This station safety risk control information can be used for event recording, personnel alerts, on-site verification, or subsequent safety management within the station safety monitoring platform.

[0134] In one implementation, the station safety risk control information may include at least the target personnel identifier, corresponding work badge identifier, event time, event location, and candidate violation event type. It may also record whether the personnel attribution result has undergone joint verification using visual and work badge trajectories, so as to facilitate subsequent tracing of the event handling process.

[0135] Through the above processing, a complete processing chain can be formed, from visual trajectory misconnection identification and personnel correspondence correction to candidate violation event personnel attribution verification.

[0136] The methods provided in this application can be applied to various scenarios, including but not limited to: personnel safety monitoring in station platforms, shunting operation areas, and equipment maintenance areas. For example, in scenarios with dense staff traffic on platforms, when multiple personnel overlap or obstruct each other, causing misalignment of personnel identification in visual tracks, the personnel correspondence can be corrected by combining the positioning trajectory of smart work badges; in shunting operation areas, personnel attribution verification can be performed on candidate violations such as entering restricted work areas or crossing safety boundaries, reducing false alarms caused by visual tracking errors; in equipment maintenance areas, the identity of personnel entering specific equipment areas or hazardous work areas can be continuously verified by combining the visual trajectory of maintenance personnel with work badge trajectory, providing reliable data support for station safety risk control, event tracing, and on-site management.

[0137] According to another embodiment, a vision-based joint verification system for employee badge trajectories is provided. Figure 4 A schematic block diagram of the vision and employee badge trajectory joint verification system according to one embodiment is shown. Figure 4 As shown, the system includes:

[0138] The trajectory acquisition module 401 is used to acquire the visual motion trajectory of multiple target personnel within the station area and the corresponding work badge positioning trajectory, and convert them to a unified coordinate system.

[0139] The common offset determination module 402 is used to determine the common trajectory offset of the corresponding area based on the positional offset between the visual motion trajectory of multiple target personnel and the work badge positioning trajectory, and to correct the work badge positioning trajectory based on the common trajectory offset.

[0140] The residual deviation determination module 403 is used to determine the individual residual deviation between the corrected name tag positioning trajectory and the corresponding visual motion trajectory.

[0141] The deviation exchange judgment module 404 is used to compare the individual residual deviation changes of each visual motion trajectory relative to the two corresponding work badge positioning trajectories before and after the intersection when two visual motion trajectories intersect, and to determine whether a deviation exchange relationship is formed.

[0142] The joint verification module 405 is used to correct the personnel correspondence of the visual motion trajectory according to the deviation exchange relationship, and to perform personnel attribution verification on the candidate violation event according to the corrected visual motion trajectory, and generate station safety risk control information.

[0143] As an implementable approach, the common offset determination module 402 determines the common trajectory offset of a corresponding area based on the positional offset between the visual motion trajectories of multiple target personnel and the nameplate positioning trajectory. This includes: for each target personnel, obtaining multiple visual positions from the visual motion trajectory according to the sampling time, and obtaining the nameplate position corresponding to the sampling time of each visual position from the corresponding nameplate positioning trajectory; determining the nameplate positioning offset direction and the nameplate positioning offset amount of the corresponding nameplate position relative to the visual position, respectively, using each visual position as a reference; comparing the nameplate positioning offsets corresponding to multiple target personnel, and determining the nameplate positioning offsets with consistent nameplate positioning offset directions and nameplate positioning offset amount differences that meet preset offset conditions as common nameplate positioning offsets; and determining the common trajectory offset of the nameplate positioning trajectory relative to the visual motion trajectory within the corresponding area based on the offset direction and offset amount of the common nameplate positioning offset.

[0144] As an implementable approach, the deviation exchange judgment module 404 determines whether a deviation exchange relationship is formed, including: identifying the two intersecting visual motion trajectories as the first visual motion trajectory and the second visual motion trajectory, respectively; identifying the badge positioning trajectories that correspond to the first and second visual motion trajectories before the intersection as the first badge positioning trajectory and the second badge positioning trajectory, respectively; determining the residual deviation of the first visual motion trajectory relative to the first and second badge positioning trajectories before the intersection, and the residual deviation of the second visual motion trajectory relative to the first and second badge positioning trajectories; determining the residual deviation of the first visual motion trajectory relative to the first and second badge positioning trajectories after the intersection, and the residual deviation of the second visual motion trajectory relative to the first and second badge positioning trajectories; comparing the changes in residual deviation before and after the intersection; and determining that the deviation exchange relationship is formed when the residual deviation of the first visual motion trajectory relative to the first badge positioning trajectory increases and the residual deviation relative to the second badge positioning trajectory decreases, and the residual deviation of the second visual motion trajectory relative to the second badge positioning trajectory increases and the residual deviation relative to the first badge positioning trajectory decreases.

[0145] As an implementable approach, before the common offset determination module 402 determines the common work badge positioning offset, it further includes: for multiple target personnel located in the same station area, determining the first relative displacement between the corresponding visual positions of any two target personnel and the second relative displacement between the corresponding work badge positions; comparing the first relative displacement and the second relative displacement corresponding to the same combination of target personnel to determine the relative displacement difference; determining the target personnel whose relative displacement differences corresponding to multiple combinations of target personnel all meet the preset consistency condition as the common offset reference personnel; and determining the common work badge positioning offset based on the work badge positioning offset corresponding to the common offset reference personnel.

[0146] As an implementable method, the common offset determination module 402 determines the common trajectory offset of the work badge positioning trajectory relative to the visual motion trajectory within the corresponding area based on the offset direction and offset amount of the common work badge positioning offset, including: converting each of the common work badge positioning offsets into work badge positioning offset vectors according to the corresponding offset direction and offset amount; determining the common offset components of the multiple work badge positioning offset vectors in each coordinate direction of the station area coordinate system; generating a common offset vector based on the common offset components in each coordinate direction; and determining the common offset vector as the common trajectory offset of the corresponding area.

[0147] As an implementable approach, the common offset determination module 402 corrects the positioning trajectory of each work badge based on the common trajectory offset, including: according to the offset direction and offset amount corresponding to the common trajectory offset, performing reverse compensation on the trajectory position of each work badge positioning trajectory in the corresponding area to obtain the corrected work badge positioning trajectory.

[0148] As an implementable approach, the deviation exchange judgment module 404 determines that when two visual motion trajectories intersect, the distance between the two visual motion trajectories is reduced to within a preset intersection distance, and then increased again to above the preset intersection distance in subsequent continuous monitoring images.

[0149] As an implementable method, the joint verification module 405 corrects the personnel correspondence of the visual motion trajectory according to the deviation exchange relationship, including: exchanging the personnel identifiers of the two visual motion trajectories that have the deviation exchange relationship after they intersect, and extending the corrected personnel identifiers to the subsequent trajectory segments of the corresponding visual motion trajectory.

[0150] As an implementable approach, the joint verification module 405 performs personnel attribution verification on candidate violation events based on the corrected visual motion trajectory, including: determining the time and location of the event corresponding to the candidate violation event; determining the target trajectory segment passing through the location of the event from the corrected visual motion trajectory; and determining the target personnel corresponding to the candidate violation event based on the personnel identification and work badge corresponding to the target trajectory segment.

[0151] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for joint verification of visual and employee badge trajectories, characterized in that, include: Acquire the visual motion trajectories and corresponding work badge positioning trajectories of multiple target personnel within the station area, and convert them to a unified coordinate system; Based on the positional offset between the visual motion trajectories of multiple target personnel and the work badge positioning trajectory, the common trajectory offset of the corresponding area is determined, and the positioning trajectory of each work badge is corrected based on the common trajectory offset. Determine the individual residual deviation between the corrected employee badge positioning trajectory and the corresponding visual motion trajectory; When two visual motion trajectories intersect, compare the individual residual deviation changes of each visual motion trajectory relative to the two corresponding work badge positioning trajectories before and after the intersection to determine whether a deviation exchange relationship has been formed. The personnel correspondence of the visual motion trajectory is corrected according to the deviation exchange relationship, and the personnel attribution of candidate violation events is verified according to the corrected visual motion trajectory to generate station safety risk control information.

2. The method according to claim 1, characterized in that, The step of determining the common trajectory offset of the corresponding area based on the positional offset between the visual motion trajectories of multiple target personnel and the work badge positioning trajectory includes: For each target person, multiple visual positions are obtained from the visual motion trajectory according to the sampling time, and the nameplate position corresponding to the sampling time of each visual position is obtained from the corresponding nameplate positioning trajectory. Using each of the aforementioned visual positions as a reference, determine the direction and amount of the badge positioning offset relative to the visual position. Compare the badge positioning offsets of multiple target personnel, and determine the badge positioning offsets that have the same positioning offset direction and whose positioning offset difference meets the preset offset conditions as common badge positioning offsets. Based on the offset direction and offset amount of the common work badge positioning offset, the common trajectory offset of the work badge positioning trajectory relative to the visual motion trajectory within the corresponding area is determined.

3. The method according to claim 1, characterized in that, Determining whether a deviation exchange relationship has been formed includes: The two visual motion trajectories that intersect are respectively defined as the first visual motion trajectory and the second visual motion trajectory, and the nameplate positioning trajectory that corresponds to the first visual motion trajectory and the second visual motion trajectory before the intersection is defined as the first nameplate positioning trajectory and the second nameplate positioning trajectory. The residual deviations of the first visual motion trajectory relative to the first and second work badge positioning trajectories before the intersection are determined, as well as the residual deviations of the second visual motion trajectory relative to the first and second work badge positioning trajectories. The residual deviations of the first visual motion trajectory relative to the first and second work badge positioning trajectories after the intersection are determined, as well as the residual deviations of the second visual motion trajectory relative to the first and second work badge positioning trajectories. By comparing the changes in residual deviations before and after the intersection, when the residual deviation of the first visual motion trajectory relative to the first work badge positioning trajectory increases and the residual deviation relative to the second work badge positioning trajectory decreases, and the residual deviation of the second visual motion trajectory relative to the second work badge positioning trajectory increases and the residual deviation relative to the first work badge positioning trajectory decreases, the deviation exchange relationship is determined to be formed.

4. The method according to claim 2, characterized in that, Before determining the common work badge positioning offset, the method further includes: For multiple target personnel located in the same station area, determine the first relative displacement between the corresponding visual positions of any two target personnel, and the second relative displacement between the corresponding name tag positions; Compare the first and second relative displacements corresponding to the same group of personnel to determine the difference in relative displacement. The target personnel whose relative displacement differences for multiple target personnel combinations all meet the preset consistency conditions are identified as common offset reference personnel. The common work badge positioning offset is determined based on the work badge positioning offset corresponding to the common offset reference personnel.

5. The method according to claim 2, characterized in that, The step of determining the common trajectory offset of the employee badge positioning trajectory relative to the visual motion trajectory within the corresponding area based on the offset direction and offset amount of the common employee badge positioning offset includes: The common work badge positioning offsets are converted into work badge positioning offset vectors according to the corresponding offset direction and offset amount; The common offset components of the multiple work badge positioning offset vectors in each coordinate direction of the station area coordinate system are determined respectively; A common offset vector is generated based on the common offset components in each coordinate direction, and the common offset vector is determined as the common trajectory offset of the corresponding region.

6. The method according to claim 1, characterized in that, The correction of each employee badge's positioning trajectory based on the common trajectory offset includes: Based on the offset direction and offset amount corresponding to the common trajectory offset, the trajectory position of each work badge positioning trajectory in the corresponding area is reversed to obtain the corrected work badge positioning trajectory.

7. The method according to claim 1, characterized in that, The two visual motion trajectories intersect, including: The distance between the two visual motion trajectories is reduced to within a preset intersection distance, and then increased again to above the preset intersection distance in subsequent continuous monitoring images.

8. The method according to claim 1, characterized in that, The step of correcting the personnel correspondence of visual motion trajectories based on the deviation exchange relationship includes: The personnel identifiers of the two visual motion trajectories that have the aforementioned deviation exchange relationship are exchanged after they intersect, and the corrected personnel identifiers are continued to the subsequent trajectory segments of the corresponding visual motion trajectory.

9. The method according to claim 1, characterized in that, The step of verifying the attribution of personnel for candidate violations based on the corrected visual motion trajectory includes: Determine the time and location of the candidate traffic violation incident; The target trajectory segment passing through the location of the event is determined from the corrected visual motion trajectory, and the target personnel corresponding to the candidate violation event are determined based on the personnel identification and work badge corresponding to the target trajectory segment.

10. A vision-based and employee badge trajectory joint verification system, characterized in that, include: The trajectory acquisition module is used to acquire the visual motion trajectories of multiple target personnel within the station area and their corresponding work badge positioning trajectories, and convert them to a unified coordinate system; The common offset determination module is used to determine the common trajectory offset of the corresponding area based on the positional offset between the visual motion trajectory of multiple target personnel and the work badge positioning trajectory, and to correct the work badge positioning trajectory based on the common trajectory offset. The residual deviation determination module is used to determine the individual residual deviation between the corrected name tag positioning trajectory and the corresponding visual motion trajectory; The deviation exchange judgment module is used to compare the individual residual deviation changes of each visual motion trajectory relative to the two corresponding work badge positioning trajectories before and after the intersection when two visual motion trajectories intersect, and to determine whether a deviation exchange relationship has been formed. The joint verification module is used to correct the personnel correspondence of visual motion trajectories according to the deviation exchange relationship, and to perform personnel attribution verification on candidate violation events according to the corrected visual motion trajectories, thereby generating station safety risk control information.