Intelligent parking system and vehicle parking management method

Abstract

The invention discloses an intelligent parking system and a vehicle parking management method, and relates to the technical field of intelligent traffic and Internet of Things, and the method comprises the steps: S1, collecting vehicle position signals and vehicle image features, and carrying out the data fusion, and obtaining the real-time coordinates of a vehicle with direction information; s2, constructing a continuous track chain based on the real-time coordinates of the vehicle, and recognizing abnormal behaviors of the vehicle based on a preset track feature threshold value; and S3, generating and executing a corresponding barrier gate control instruction according to the abnormal behavior of the vehicle. According to the method, geomagnetism and video sensing data are fused in a dynamic self-adaptive mode, a vehicle continuous track chain is constructed, abnormal behaviors such as vehicle retention and turning around can be accurately recognized, real-time linkage with barrier gate control is achieved, and therefore the vehicle smashing prevention safety, the accuracy of illegal behavior recognition and the real-time performance of system response are effectively improved.

Description

Technical Field

[0001] This invention relates to the fields of intelligent transportation and the Internet of Things, specifically to a smart parking system and a vehicle parking management method. Background Technology

[0002] With the acceleration of urbanization and the continuous growth of motor vehicle ownership, unmanned parking lots have become the mainstream form of smart parking. Currently, the industry generally adopts a solution combining geomagnetic detectors and video sensors to achieve basic functions such as vehicle detection, license plate recognition, automatic billing, and barrier gate linkage. These systems typically operate based on a "single-sensor trigger simple control" model: for example, at the entrance, the system raises the barrier gate after recognizing the license plate via video; at the exit, the barrier gate lowers after billing is completed; and anti-collision protection mainly relies on a single infrared sensor integrated into the barrier gate itself.

[0003] However, existing technological solutions still have significant limitations. For example, they only focus on basic parking space occupancy detection and license plate recognition, failing to form a closed-loop management system for vehicle trajectories. Furthermore, due to isolated sensing data and rudimentary control logic, existing solutions still have technical shortcomings in core scenarios such as anti-collision vehicle security and illegal U-turn detection at entrances, making it difficult to balance management efficiency and safety reliability. Moreover, existing technologies are mostly conventional integrations of mature components, lacking targeted technological innovation and failing to meet the actual needs of high-security unattended scenarios.

[0004] Therefore, the reliability, accuracy, and real-time performance of existing technical solutions need to be improved when dealing with unattended scenarios with high security and order requirements. Summary of the Invention

[0005] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide an intelligent parking system and a vehicle parking management method to solve the above-mentioned technical problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a smart parking system and a vehicle parking management method, comprising: S1: Collect vehicle position coordinate sequence and image coordinate sequence, perform spatiotemporal alignment on the coordinate sequence, and dynamically calculate fusion weight based on the fluctuation of the vehicle position coordinate sequence and the recognition confidence of the image. Based on the fusion weight, perform weighted fusion on the aligned coordinates to generate real-time vehicle coordinates with directional information. S2: Based on the real-time coordinates of the vehicle, a trajectory chain composed of continuous trajectory points is constructed in chronological order. By analyzing whether the position changes and orientation angle changes of the trajectory points in the trajectory chain meet the preset trajectory feature thresholds, abnormal vehicle behavior can be identified. S3: Generate and execute corresponding gate control commands based on abnormal vehicle behavior.

[0007] The present invention is further configured such that S1 includes: a data acquisition and processing step, a fusion weight calculation step, and a data fusion and direction calculation step.

[0008] The present invention is further configured such that the data acquisition and processing steps include: The system uses a geomagnetic detector to collect vehicle position coordinate sequences at a preset first time interval, and uses video stakes to collect image coordinate sequences containing vehicle feature points at a preset second time interval. Align the vehicle location coordinate sequence with the image coordinate sequence based on timestamps and convert them to a unified preset parking lot terrain coordinate system; The degree of fluctuation of the geomagnetic signal is determined by calculating the coordinate difference of continuously collected data in the vehicle position coordinate sequence, and the recognition confidence is determined based on the vehicle detection model and the recognition results of vehicle feature points in the image coordinate sequence.

[0009] The present invention is further configured such that the fusion weight calculation step includes: The fluctuation level of the geomagnetic signal is converted into a geomagnetic quality score according to a preset rule; The confidence level of video recognition is used as the video quality score; Calculate the geometric mean of the geomagnetic quality score and the video quality score to obtain a confidence factor that characterizes the current data fusion quality; The credibility factor is compared with a preset threshold, and based on the comparison results, a fusion weight allocation scheme that prioritizes geomagnetic data, video data, or a balance between the two is dynamically selected.

[0010] The present invention is further configured such that the data fusion and direction calculation steps include: Based on the fusion weights determined by the fusion weight allocation scheme, the aligned vehicle position coordinates and image coordinates from the same moment are fused and calculated to generate real-time vehicle coordinates. The real-time driving direction angle of a vehicle is determined based on the changes in the real-time coordinates of multiple vehicles.

[0011] The present invention is further configured such that the abnormal vehicle behavior includes at least: vehicle lingering and U-turn at the entrance; The S2 includes: a trajectory chain construction step, an abnormal behavior determination step, and an identification result generation step.

[0012] The present invention is further configured such that the trajectory chain construction step includes: Based on the vehicle's real-time coordinates, they are sorted according to timestamps to form a coordinate sequence arranged in chronological order; Based on a preset sampling interval or a preset number of coordinate points, a continuous set of trajectory points is extracted from the coordinate sequence to form a trajectory chain that represents the continuous movement path of the vehicle.

[0013] The present invention is further configured such that the abnormal behavior determination step includes: Based on the trajectory chain, determine whether the vehicle is within the preset gate control area; When a vehicle is within the gate control area, the position and direction angle changes of multiple consecutive trajectory points in the trajectory chain are analyzed to determine whether the vehicle is lingering. When a vehicle is in the entrance passage area, the system analyzes the trend of the change in the driving direction angle and the direction of coordinate movement of the trajectory chain within a preset time window to determine whether the vehicle is making a U-turn.

[0014] The present invention is further configured such that the step of generating the recognition result includes: Based on the results of the abnormal behavior determination steps, the abnormal behavior type, the time window information of the determination basis, the triggering threshold conditions, the timestamp of the abnormal occurrence, and the associated vehicle identification information are encapsulated into a structured vehicle abnormal behavior event record.

[0015] The present invention is further configured such that S3 includes: Based on the identification results and types of abnormal vehicle behavior, gate control commands are generated according to the preset corresponding control strategies. The gate control command is sent to the gate controller to execute the corresponding gate opening / closing or status maintenance operation.

[0016] This invention provides a smart parking system and a vehicle parking management method. The method comprises: S1: collecting vehicle position coordinate sequences and image coordinate sequences; spatiotemporally aligning the coordinate sequences; dynamically calculating fusion weights based on the fluctuation of the vehicle position coordinate sequences and the recognition confidence of the image; and weighted fusion of the aligned coordinates based on these fusion weights to generate real-time vehicle coordinates with directional information; S2: constructing a trajectory chain composed of continuous trajectory points in chronological order based on the real-time vehicle coordinates; and identifying abnormal vehicle behavior by analyzing whether the position changes and directional angle changes of the trajectory points in the trajectory chain meet preset trajectory feature thresholds; and S3: generating and executing corresponding gate control commands based on the abnormal vehicle behavior. The beneficial effects include: Dynamic adaptive multi-source perception fusion method: Before data fusion, the fluctuation degree of geomagnetic signal and the confidence degree of video recognition are calculated to generate a "confidence factor". Based on this factor, a fusion weight allocation scheme dominated by geomagnetic data or video data is dynamically selected to improve the accuracy of vehicle state perception and environmental adaptability.

[0017] A precise abnormal behavior identification method based on continuous trajectory chains: By constructing a trajectory chain of continuous vehicle movement and setting specific geometric and kinematic thresholds (such as changes in direction angle and coordinate backtracking distance), the method can model and quantify the identification and judgment of two specific abnormal behaviors: "vehicle stagnation" and "entrance U-turn".

[0018] Real-time barrier gate linkage control mechanism for abnormal behavior: The system directly triggers the pre-set corresponding barrier gate control strategy (such as stopping the descent, refusing to open, or maintaining the state) based on the specific abnormal behavior type identified (such as lingering or turning around), realizing a real-time and precise closed loop from behavior perception to safety control command.

[0019] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 The flowchart illustrates an intelligent parking system and vehicle parking management method as an exemplary embodiment of the present invention. Detailed Implementation

[0021] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0022] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0023] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0024] Example: A smart parking system and vehicle parking management method, such as Figure 1 As shown, it includes: S1: Collect vehicle position coordinate sequence and image coordinate sequence, perform spatiotemporal alignment on the coordinate sequence, and dynamically calculate fusion weight based on the fluctuation of the vehicle position coordinate sequence and the recognition confidence of the image. Based on the fusion weight, perform weighted fusion on the aligned coordinates to generate real-time vehicle coordinates with directional information. S2: Based on the real-time coordinates of the vehicle, a trajectory chain composed of continuous trajectory points is constructed in chronological order. By analyzing whether the position changes and orientation angle changes of the trajectory points in the trajectory chain meet the preset trajectory feature thresholds, abnormal vehicle behavior can be identified. S3: Generate and execute corresponding gate control commands based on abnormal vehicle behavior.

[0025] The present invention is further configured such that S1 includes: a data acquisition and processing step, a fusion weight calculation step, and a data fusion and direction calculation step. Specifically, the data acquisition and processing step is used to acquire the raw signals from the geomagnetic detector and the video stake, and convert the raw signals into unified coordinate data aligned in time and space to provide standardized input for fusion; the fusion weight calculation step is used to dynamically determine the weight allocation of geomagnetic data and video data in the fusion process based on the fluctuation degree of the geomagnetic signal and the confidence level of video recognition, thereby realizing adaptive adjustment of the fusion strategy according to data quality; the data fusion and direction calculation step is used to perform weighted calculation on the coordinate data according to the fusion weight determined in the previous step, generate real-time vehicle coordinates, and calculate the vehicle driving direction angle based on the continuous coordinates, thereby outputting high-precision vehicle pose data with direction information.

[0026] The present invention is further configured such that the data acquisition and processing steps include: The system uses a geomagnetic detector to collect vehicle position coordinate sequences at a preset first time interval, and uses video stakes to collect image coordinate sequences containing vehicle feature points at a preset second time interval. Align the vehicle location coordinate sequence with the image coordinate sequence based on timestamps and convert them to a unified preset parking lot terrain coordinate system; The dynamism of the geomagnetic signal is determined by calculating the coordinate difference of continuously acquired data in the vehicle position coordinate sequence, and the recognition confidence is determined based on the vehicle feature point recognition results in the image coordinate sequence by the vehicle detection model. Specifically, a geomagnetic detector is used to acquire vehicle position coordinate sequences at a preset first time interval (e.g., 100 milliseconds). Each coordinate is converted from a real-time voltage value to a two-dimensional coordinate in the global terrain coordinate system of the parking lot through a pre-stored voltage-position mapping table. Simultaneously, video stakes are used to acquire images at a preset second time interval (e.g., 40 milliseconds). For each frame of the image, the YOLOv8 vehicle detection model is used to identify the vehicle bounding box, and three feature points are located within the box: the center point of the front of the vehicle, the center point of the rear of the vehicle, and the highest point of the roof of the vehicle. Through the internal parameters, distortion coefficients, and external rigid body transformation matrix relative to the global terrain coordinate system of the parking lot obtained by camera calibration, the pixel coordinates of these feature points are converted to coordinates in the same global terrain coordinate system. Then, spatiotemporal alignment is performed. Using the system clock as a reference, a linear interpolation method is used to synchronize the geomagnetic coordinates and video feature point coordinates to the same time point, ensuring that each aligned time point contains a set of geomagnetic coordinates and video feature point coordinates. Next, data quality indicators are calculated: the degree of fluctuation is evaluated by calculating the Euclidean distance difference between the geomagnetic coordinates at three consecutive aligned times. A fluctuation threshold is set (e.g., 0.02 meters). If two consecutive distance differences do not exceed 0.02 meters, the geomagnetic signal is considered stable; otherwise, if any distance difference exceeds the threshold, it is considered fluctuating. The recognition confidence score is obtained by multiplying the vehicle detection confidence score output by the YOLOv8 model with the quality score derived from feature point image sharpness analysis. The resulting recognition confidence score is between 0 and 1. Finally, this step outputs geomagnetic coordinates and video feature point coordinates aligned at a unified time point and in a consistent coordinate system. It also outputs the geomagnetic signal fluctuation degree determination result (stable or fluctuating) based on the distance difference comparison, as well as the video recognition confidence score.

[0027] The present invention is further configured such that the fusion weight calculation step includes: The fluctuation level of the geomagnetic signal is converted into a geomagnetic quality score according to a preset rule; The confidence level of video recognition is used as the video quality score; Calculate the geometric mean of the geomagnetic quality score and the video quality score to obtain a confidence factor that characterizes the current data fusion quality; The confidence factor is compared with a preset threshold, and based on the comparison result, a dynamic selection of a fusion weight allocation scheme is made, which prioritizes geomagnetic data, video data, or a balance between the two. Specifically, first, the fluctuation degree judgment result and recognition confidence value output from the data acquisition and processing steps are received. The fluctuation degree judgment result is a Boolean value, taking the value of "stable" or "fluctuating," while the video recognition confidence value is a floating-point number between 0 and 1. Second, the confidence factor is calculated: the Boolean fluctuation degree judgment result is converted into a specific geomagnetic quality score. The rule is that if the judgment is "stable," the geomagnetic quality score is 1.0, and if the judgment is "fluctuating," the geomagnetic quality score is 0.2. This score reflects the relative confidence of the geomagnetic data; a higher score indicates stability, and a lower score indicates fluctuation. The video recognition confidence value is directly used as the video quality score. The geometric mean method is used to combine these two scores, that is, the product of the geomagnetic quality score and the video quality score is calculated, and then the square root of the product is taken. The result is the confidence factor, which takes the value between 0 and 1. Subsequently, a dynamic fusion weight allocation scheme is selected: the system presets two thresholds, for example, a high threshold of 0.7 and a low threshold of 0.3; then, the calculated confidence factor is compared with these two thresholds. If the confidence factor is greater than or equal to 0.7, a geomagnetic-dominated fusion weight allocation scheme is selected, for example, assigning a weight of 0.7 to geomagnetic data and a weight of 0.3 to video data; if the confidence factor is less than 0.7 but greater than or equal to 0.3, a balanced fusion weight allocation scheme is selected, assigning a weight of 0.5 to both geomagnetic data and video data; if the confidence factor is less than 0.3, a video-dominated fusion weight allocation scheme is selected, for example, assigning a weight of 0.3 to geomagnetic data and a weight of 0.7 to video data. Finally, this fusion weight calculation step outputs the specific fusion weight values ​​corresponding to the finally selected fusion weight allocation scheme.

[0028] The present invention is further configured such that the data fusion and direction calculation steps include: Based on the fusion weights determined by the fusion weight allocation scheme, the aligned vehicle position coordinates and image coordinates from the same moment are fused and calculated to generate real-time vehicle coordinates. The real-time driving direction angle of a vehicle is determined based on the changes in the real-time coordinates of multiple vehicles. Specifically, based on the fusion weights determined by the fusion weight allocation scheme, the aligned vehicle position coordinates and image coordinates from the same moment are fused to generate the real-time vehicle coordinates. The fusion weights include geomagnetic data weights and video data weights, the sum of which is one. The aligned vehicle position coordinates are those provided by the geomagnetic detector, and the aligned image coordinates are those of the vehicle's center point provided by the video stake. The fusion calculation uses a weighted average algorithm: the horizontal coordinate of the geomagnetic coordinates is multiplied by the geomagnetic data weight, and then the horizontal coordinate of the image coordinates is multiplied by the video data weight to obtain the fused horizontal coordinate; the vertical coordinate of the geomagnetic coordinates is multiplied by the geomagnetic data weight, and then the vertical coordinate of the image coordinates is multiplied by the video data weight to obtain the fused vertical coordinate. Subsequently, the real-time driving direction angle of the vehicle is determined based on the changes in the real-time coordinates of multiple vehicles. During the calculation, the fused coordinates at the current moment and the fused coordinates cached at the previous moment are taken to calculate the eastward and northward displacements. The eastward displacement is equal to the current x-coordinate minus the x-coordinate at the previous moment, and the northward displacement is equal to the current y-coordinate minus the y-coordinate at the previous moment. Using the four-quadrant arctangent function, the northward and eastward displacements are input as parameters to calculate the radian value of the angle between the displacement vector and the due east direction. This radian value is multiplied by 180 and divided by pi (π) to convert it into an angle value. This angle value is substituted into the formula "90 degrees minus this angle value" for calculation. If the result is less than 0 degrees, 360 degrees are added. Finally, the real-time driving direction angle measured clockwise with due north as 0 degrees is obtained, and its value range is from 0 degrees to 360 degrees. Finally, the coordinate cache is updated, and the fused real-time vehicle coordinates and their corresponding real-time driving direction angles are output, which together constitute the vehicle pose data with directional information.

[0029] The present invention is further configured such that the abnormal vehicle behavior includes at least: vehicle lingering and U-turn at the entrance; S2 includes: a trajectory chain construction step, an abnormal behavior determination step, and an identification result generation step. Specifically, firstly, abnormal behavior refers to the vehicle driving state that deviates from the preset normal traffic pattern within a specific control area of ​​the parking lot. In this embodiment, abnormal vehicle behaviors include at least: vehicle lingering and entrance U-turn. Among them, vehicle lingering refers to a vehicle lingering for a long time under or near the preset control area of ​​the barrier gate. The specific determination condition is that the vehicle's position coordinates have not changed significantly for a long time, which is consistent with the departure trend. The purpose of setting this abnormal behavior is to serve as the core detection target of the anti-collision safety protection mechanism, so as to accurately identify this behavior and trigger the barrier gate to stop falling or reopen, thereby fundamentally avoiding physical collision accidents. Entrance U-turn refers to the behavior of a vehicle in the parking lot entrance channel that has not completed the normal entry process and has changed its driving direction angle by nearly 180 degrees and moved towards the entrance. The purpose of setting this abnormal behavior is to serve as a key detection target for traffic order control, so as to accurately identify this behavior and prevent the barrier gate from being opened erroneously, avoid duplicate entry records, and maintain the correctness of the billing system and the smoothness of entrance traffic. Secondly, step S2 specifically includes: a trajectory chain construction step, an abnormal behavior determination step, and an identification result generation step. The trajectory chain construction step refers to organizing continuously acquired, timestamped real-time vehicle coordinates into a data sequence representing the continuous spatiotemporal path of a vehicle, providing a structured temporal data foundation for subsequent behavior analysis and transforming discrete location points into a complete trajectory capable of analyzing their movement trends, speed, and direction changes. The abnormal behavior determination step refers to using the constructed trajectory chain data and applying preset geometric and kinematic feature thresholds (such as coordinate region, change in direction angle, and displacement) to logically identify specific behaviors such as vehicle stagnation or U-turns at entrances. This step transforms raw trajectory data into specific behavior category judgments and is the core of the system's intelligent decision-making for precise control. The identification result generation step encapsulates the logical judgment results generated in the abnormal behavior determination step into standardized data records containing information such as behavior type, judgment basis, timestamp, and associated vehicle identifiers. This provides clear, structured, and processable instruction or event inputs for subsequent gate control, system alarms, and data traceability, enabling effective linkage between various modules of the system.

[0030] The present invention is further configured such that the trajectory chain construction step includes: Based on the vehicle's real-time coordinates, they are sorted according to timestamps to form a coordinate sequence arranged in chronological order; Based on a preset sampling interval or a preset number of coordinate points, a continuous set of trajectory points is extracted from the coordinate sequence to form a trajectory chain representing the vehicle's continuous movement path. Specifically, the vehicle's real-time coordinates are data packets containing timestamps, position coordinates, and orientation angles. First, the system stores all received vehicle real-time coordinate data packets into a data buffer that is automatically sorted by timestamp, ensuring that the output is a coordinate sequence strictly arranged in ascending order of timestamps. Then, based on a preset trajectory sampling interval (e.g., 100 milliseconds) and a preset trajectory chain capacity (e.g., 50 points), a continuous set of trajectory points is extracted from the coordinate sequence to form a trajectory chain representing the vehicle's continuous movement path. The system maintains a fixed-capacity FIFO queue of 50 points to store trajectory chains. When extracting trajectory points from the coordinate sequence, target times are sampled at 100-millisecond intervals. For each target sampling time, the system searches the coordinate sequence for the coordinate point whose timestamp is closest to that time. If no point's timestamp is exactly the same as the target time, an interpolation algorithm is used to generate trajectory points: the algorithm first finds the two closest coordinate points before and after the target time in the coordinate sequence, and then calculates the estimated values ​​of the x-coordinate and y-coordinate of the target time based on the position ratio of the target time within the time interval between these two points. For the orientation angle, its circular cyclic characteristics need to be considered, and an angle interpolation method (such as calculating the shortest path interpolation) is used to estimate the orientation angle of the target time. The obtained trajectory points are added to the end of the queue. Whenever a new trajectory point is added, the system checks the queue length. If it exceeds 50 points, the point at the front of the queue (i.e., the earliest time) is automatically removed, ultimately forming a trajectory chain representing the continuous movement path of the vehicle. This process continues, ensuring that the trajectory chain always contains the latest 50 vehicle pose points sampled at 100-millisecond intervals, which are then used as the final output.

[0031] The present invention is further configured such that the abnormal behavior determination step includes: Based on the trajectory chain, determine whether the vehicle is within the preset gate control area; When a vehicle is within the gate control area, the position and direction angle changes of multiple consecutive trajectory points in the trajectory chain are analyzed to determine whether the vehicle is lingering. When a vehicle is in the entrance passage area, the system analyzes the trend of its driving direction angle and coordinate movement within a preset time window to determine if it is making a U-turn. Specifically, based on the trajectory chain, it determines whether the vehicle is within a preset gate control area. The trajectory chain is the output of the trajectory chain construction step and contains 50 vehicle pose points sampled at 100-millisecond intervals. Each point includes a timestamp, position coordinates, and driving direction angle. The gate control area is a pre-defined polygonal region in the parking lot's global terrain coordinate system, such as a rectangle defined by the coordinates of four vertices. The determination method uses a ray casting approach: the five most recent points are extracted from the trajectory chain; for each point, a horizontal ray is drawn to the right from that point's coordinates; the number of intersections between this ray and each side of the gate control area polygon is calculated; if the total number of intersections is odd, the point is inside the polygon; otherwise, it is outside the polygon; if all five points are determined to be within the gate control area, the vehicle is determined to be within the gate control area. When a vehicle is within the gate control area, the system analyzes the position and direction angle changes of multiple consecutive trajectory points in the trajectory chain to determine if the vehicle is lingering. To determine if a vehicle is lingering, the system extracts the five most recent consecutive points from the trajectory chain and calculates the straight-line distance between the first and fifth points. Simultaneously, to calculate the change in direction angle, the driving direction angle values ​​of these five points are first converted to a continuous angle space (for example, subtracting 360 degrees from an angle greater than 180 degrees). Then, the standard deviation of the converted angle values ​​is calculated. If the calculated straight-line distance is less than 0.2 meters and the standard deviation of the direction angle is less than 5 degrees, the system determines that the vehicle is lingering. When a vehicle is in the entrance passage area, the system analyzes the trend of the change in the driving direction angle and the direction of coordinate movement of the trajectory chain within a preset time window to determine whether the vehicle is making a U-turn. The entrance passage area is another preset polygonal area in the global terrain coordinate system of the parking lot. The determination of whether a vehicle is in this area can adopt a similar determination logic as that of the gate control area. For example, check whether the three most recent points in the trajectory chain are all in this area. If the vehicle is in the entrance passage area, a U-turn behavior determination is made. When determining the U-turn behavior, the preset time window is 3 seconds, corresponding to the 30 most recent points in the trajectory chain. The system takes the start and end points of the time window and calculates the total change in direction angle: the direction angle of the end point is subtracted from the direction angle of the start point to obtain the original difference. If the original difference is greater than 180 degrees, 360 degrees are subtracted. If the original difference is less than -180 degrees, 360 degrees are added. Then, the absolute value of the adjusted difference is taken. If the absolute value is between 170 degrees and 190 degrees, the direction change condition is met. Calculate the direction of coordinate movement: Calculate the difference in x-coordinates between the end point and the starting point. Assuming that the positive direction of the X-axis of the parking lot coordinate system points inward, then movement towards the entrance is represented by a negative x-coordinate difference. If the x-coordinate difference is negative and its absolute value is greater than 0.5 meters, then the coordinate backtracking condition is met. If both the direction change condition and the coordinate backtracking condition are met, then it is determined that the vehicle has made a U-turn.Finally, this step outputs the judgment results regarding vehicle lingering and U-turn behavior.

[0032] The present invention is further configured such that the step of generating the recognition result includes: According to the determination result of the abnormal behavior determination step, encapsulate the abnormal behavior type, the time window information on which the determination is based, the triggered threshold conditions, the timestamp of the abnormal occurrence, and the associated vehicle identification information into a structured vehicle abnormal behavior event record. Specifically, according to the determination result of the abnormal behavior determination step, encapsulate the abnormal behavior type, the time window information on which the determination is based, the triggered threshold conditions, the timestamp of the abnormal occurrence, and the associated vehicle identification information into a structured vehicle abnormal behavior event record. When the abnormal behavior determination step outputs a result indicating an abnormality, this step is initiated. First, generate a globally unique event identifier, for example: generate a string using the UUID version 4 algorithm; then, map the abnormal type in the determination result to a standard event type string, for example, vehicle stay is mapped to "VEHICLE_STAY", and entry U-turn is mapped to "ENTRY_U_TURN"; then, process the time information: extract the end timestamp of the time window used as the determination basis from the determination result, which is a Unix timestamp in milliseconds; use the standard time library to convert this timestamp into a date and time format string following the ISO 8601 standard as the value of the event occurrence time field; at the same time, convert both the start timestamp and the end timestamp of this time window into ISO 8601 format strings to jointly form the analysis window time field; after that, encapsulate the determination basis: extract a list from the determination result, and this list details the names and specific values of the threshold conditions that were satisfied during the determination process and led to the establishment of the abnormality. For example, for the stay behavior, the list includes "position change threshold 0.2 meters" and "direction angle change threshold 5 degrees", and this list is directly used as the value of the trigger condition field; next, associate the vehicle identification: query the session context established at the time of vehicle entry and bound to the current abnormal trajectory chain, and obtain the license plate number string that has been recognized and verified through video, such as "Beijing A12345", as the value of the vehicle identification field; if the license plate is unavailable, fill in the temporary identifier generated by the system for this vehicle session. In addition, intercept the original trajectory point data corresponding exactly to the determination time window from the trajectory chain, including the timestamp, position coordinates, and driving direction angle of each point, as the value of the trajectory snapshot field. Finally, combine all the above fields - event identifier, event type string, event occurrence time string, analysis window time object, trigger condition list, vehicle identification string, trajectory snapshot array - into a dictionary object, and use the JSON serialization library to encode this dictionary object into a string conforming to the JSON specification. This string is the finally generated structured vehicle abnormal behavior event record. This vehicle abnormal behavior event record is published to the topic named "alarm / event" through the MQTT protocol and synchronously written into the relational database log table of the system.

[0033] The present invention is further configured such that S3 includes: Based on the identification results and types of abnormal vehicle behavior, gate control commands are generated according to the preset corresponding control strategies. The system sends the gate control command to the gate controller to execute the corresponding gate opening / closing or status maintenance operation. Specifically, based on the identification result and type of abnormal vehicle behavior, a gate control command is generated according to the preset corresponding control strategy, and the gate control command is sent to the gate controller to execute the corresponding gate opening / closing or status maintenance operation. During system initialization, the preset gate control strategy configuration table is loaded into memory. This configuration table defines the mapping relationship between abnormal behavior types and specific control actions. For example, the vehicle lingering type is mapped to the "stop first, then open" action, and the entrance U-turn type is mapped to the "keep closed" action. A target gate identifier and command priority are specified for each action. The system simultaneously starts a message queue client and subscribes to a message topic named "alarm / event". When the identification result generation step publishes a structured vehicle abnormal behavior event record to this topic, the control step is triggered. The system receives the event record, which is a JSON format string. The system uses a JSON parsing library to parse the string and extract the event type field and vehicle identifier field. The system queries the loaded control strategy configuration table using the event type as the key to obtain the corresponding control action, target gate identifier, and priority information. For example, if the query finds a control action of "stop first, then open", a target gate identifier of "entrance gate one", and a priority of "high", and the target gate identifier is "automatic selection", the system automatically determines the gate area where the vehicle is located based on the location information or associated trajectory data attached to the event record, thereby determining the specific target gate controller identifier. Next, the system encapsulates the logical control action into a specific physical control instruction frame according to the hardware communication protocol of the target gate controller. For example, if the target gate controller uses the Modbus RTU protocol and the address is 1, the "stop first, then open" action is converted into two consecutive Modbus write register instructions: first, write the value 0x0005 (representing an emergency stop instruction) to register address 0x0002, wait 100 milliseconds, and then write the value 0x0001 (representing an open instruction) to the same register address. The system processes the instruction according to its priority. "High-priority" instructions are sent immediately, with the encapsulated instruction data frame transmitted to the target gate controller via an established communication link (such as an RS-485 serial port or TCP network socket). Upon receiving and verifying the instruction, the gate controller drives its internal motor to perform a "stop-then-open" physical action, immediately halting the current movement and reversing to raise the gate arm. After sending the instruction, the system marks the gate in a "safety intervention locked" state in its internal state machine. In this state, routine control requests from other modules (such as gate-closing instructions after billing completion) are temporarily stored or ignored until the system confirms, based on subsequent trajectory analysis, that the vehicle has safely left the controlled area, at which point the locked state is released.The system records the complete log of this control operation, including timestamp, event identifier, target gate, sent command content, and execution result, to the system database. At this point, the gate control execution steps complete a closed loop from abnormal behavior identification to physical security intervention.

[0034] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A smart parking system and vehicle parking management method, characterized in that, include: S1: Collect vehicle position coordinate sequence and image coordinate sequence, perform spatiotemporal alignment on the coordinate sequence, and dynamically calculate fusion weight based on the fluctuation of the vehicle position coordinate sequence and the recognition confidence of the image. Based on the fusion weight, perform weighted fusion on the aligned coordinates to generate real-time vehicle coordinates with directional information. S2: Based on the real-time coordinates of the vehicle, a trajectory chain composed of continuous trajectory points is constructed in chronological order. By analyzing whether the position changes and orientation angle changes of the trajectory points in the trajectory chain meet the preset trajectory feature thresholds, abnormal vehicle behavior can be identified. S3: Generate and execute corresponding gate control commands based on abnormal vehicle behavior.

2. The intelligent parking system and vehicle parking management method according to claim 1, characterized in that, S1 includes: data acquisition and processing steps, fusion weight calculation steps, and data fusion and direction calculation steps.

3. The intelligent parking system and vehicle parking management method according to claim 2, characterized in that, The data acquisition and processing steps include: The system uses a geomagnetic detector to collect vehicle position coordinate sequences at a preset first time interval, and uses video stakes to collect image coordinate sequences containing vehicle feature points at a preset second time interval. Align the vehicle location coordinate sequence with the image coordinate sequence based on timestamps and convert them to a unified preset parking lot terrain coordinate system; The degree of fluctuation of the geomagnetic signal is determined by calculating the coordinate difference of continuously collected data in the vehicle position coordinate sequence, and the recognition confidence is determined based on the vehicle detection model and the recognition results of vehicle feature points in the image coordinate sequence.

4. The intelligent parking system and vehicle parking management method according to claim 3, characterized in that, The fusion weight calculation steps include: The fluctuation level of the geomagnetic signal is converted into a geomagnetic quality score according to a preset rule; The confidence level of video recognition is used as the video quality score; Calculate the geometric mean of the geomagnetic quality score and the video quality score to obtain a confidence factor that characterizes the current data fusion quality; The credibility factor is compared with a preset threshold, and based on the comparison results, a fusion weight allocation scheme that prioritizes geomagnetic data, video data, or a balance between the two is dynamically selected.

5. The intelligent parking system and vehicle parking management method according to claim 4, characterized in that, The data fusion and direction calculation steps include: Based on the fusion weights determined by the fusion weight allocation scheme, the aligned vehicle position coordinates and image coordinates from the same moment are fused and calculated to generate real-time vehicle coordinates. The real-time driving direction angle of a vehicle is determined based on the changes in the real-time coordinates of multiple vehicles.

6. The intelligent parking system and vehicle parking management method according to claim 1, characterized in that, The abnormal vehicle behavior includes at least: vehicle lingering and U-turns at entrances; The S2 includes: a trajectory chain construction step, an abnormal behavior determination step, and an identification result generation step.

7. The intelligent parking system and vehicle parking management method according to claim 6, characterized in that, The trajectory chain construction steps include: Based on the vehicle's real-time coordinates, they are sorted according to timestamps to form a coordinate sequence arranged in chronological order; Based on a preset sampling interval or a preset number of coordinate points, a continuous set of trajectory points is extracted from the coordinate sequence to form a trajectory chain that represents the continuous movement path of the vehicle.

8. The intelligent parking system and vehicle parking management method according to claim 7, characterized in that, The abnormal behavior determination steps include: Based on the trajectory chain, determine whether the vehicle is within the preset gate control area; When a vehicle is within the gate control area, the position and direction angle changes of multiple consecutive trajectory points in the trajectory chain are analyzed to determine whether the vehicle is lingering. When a vehicle is in the entrance passage area, the system analyzes the trend of the change in the driving direction angle and the direction of coordinate movement of the trajectory chain within a preset time window to determine whether the vehicle is making a U-turn.

9. The intelligent parking system and vehicle parking management method according to claim 8, characterized in that, The steps for generating the recognition result include: Based on the results of the abnormal behavior determination steps, the abnormal behavior type, the time window information of the determination basis, the triggering threshold conditions, the timestamp of the abnormal occurrence, and the associated vehicle identification information are encapsulated into a structured vehicle abnormal behavior event record.

10. The intelligent parking system and vehicle parking management method according to claim 1, characterized in that, S3 includes: Based on the identification results and types of abnormal vehicle behavior, gate control commands are generated according to the preset corresponding control strategies. The gate control command is sent to the gate controller to execute the corresponding gate opening / closing or status maintenance operation.

Images