Vehicle illegal parking management and control method and system based on unmanned aerial vehicle

Abstract

The invention relates to a vehicle illegal parking management and control method and system based on an unmanned aerial vehicle. The method comprises the following steps: obtaining road information of a target controlled road section; arranging a flight plan of the unmanned aerial vehicle based on the road information; acquiring image information acquired when the unmanned aerial vehicle cruises according to the flight plan; and according to the image information, analyzing whether a ground vehicle of the target management and control road section violates parking, if yes, enabling the unmanned aerial vehicle to perform snapshot, then enabling the unmanned aerial vehicle to pause the current route and adjusting the pan-tilt camera to perform snapshot again so as to obtain a clear license plate image. According to the invention, the vehicle illegal parking event of the target road section can be accurately and efficiently identified, and vehicle illegal parking intelligent control based on unmanned aerial vehicle automatic cruise is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control of illegal parking of vehicles, and in particular to a method and system for controlling illegal parking of vehicles based on unmanned aerial vehicles. Background Art

[0002] Illegal parking of motor vehicles affects the image and order of the city, hinders traffic, and increases the risk of traffic accidents. The monitoring equipment "illegal parking ball" has a long installation period and a limited monitoring range. Complex road conditions and hidden illegal parking behaviors put higher demands on monitoring technology. In addition, the control of illegal parking of motor vehicles through traffic police patrols has high labor costs and low law enforcement efficiency.

[0003] Drones are efficient, flexible and precise, and are gradually being put into traffic control systems. The high precision and high speed of AI algorithms have been empowering various industries. However, there is a lack of a systematic solution that can accurately and efficiently identify illegal parking incidents on target roads and realize intelligent control of illegal parking based on automatic cruising of drones. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method that can accurately and efficiently identify vehicle illegal parking incidents on a target road section and realize intelligent control of vehicle illegal parking based on automatic cruising of unmanned aerial vehicles.

[0005] The technical solution adopted by the present invention to solve the technical problem is: to provide a vehicle illegal parking control method based on a drone, comprising the following steps:

[0006] Obtain road information of the target controlled section;

[0007] Arrange a flight plan of the drone based on the road information;

[0008] Obtaining image information collected by the drone when cruising according to the flight plan;

[0009] The image information is used to analyze whether ground vehicles on the target controlled road section have illegally parked. If illegal parking has occurred, the drone is ordered to take a snapshot, and then the drone is ordered to pause the current route and adjust the gimbal camera to take another snapshot to obtain a clear license plate image.

[0010] Furthermore, the arranging a flight plan of the drone based on the road information includes:

[0011] The route management information of the UAV is set based on the road information, wherein the route management information includes a flight state and a flight segment, wherein the flight state includes a flight altitude, a flight speed, a pitch angle, a yaw angle, a roll angle, and a focal length of a gimbal camera, and the flight segment includes a take-off point, each intermediate waypoint, a return point, and a corresponding flight state;

[0012] The cruise mission of the drone is set, and the cruise mission includes a mission type, and the mission type includes immediate execution, scheduled execution, and periodic execution.

[0013] Furthermore, before the step of acquiring the image information collected by the drone when cruising according to the flight plan, the step of calibrating the drone under the set route management information is also included.

[0014] Furthermore, the calibrating the UAV under the set route management information includes:

[0015] Establish a geographic coordinate system with the take-off point as the coordinate origin;

[0016] From the coordinate origin, surveying and calibration are performed along the X-axis and Y-axis directions at set intervals to obtain the mapping relationship between pixel coordinates and geographic coordinates.

[0017] Furthermore, the analyzing, based on the image information, whether a ground vehicle on the target controlled road section has an illegal parking incident includes:

[0018] Obtain parking space information, illegal parking areas, and illegal parking types corresponding to different illegal parking areas on the target controlled road section;

[0019] Construct a vehicle detection model based on the YOLOv8 algorithm and integrating the rotating target detection algorithm and the key point detection algorithm to perform target recognition on the image information and obtain the rotating vehicle information and the vehicle head orientation information;

[0020] If there is a parking area on the road section, whether illegal parking has occurred is determined based on the vehicle's front direction information, vehicle position information and parking space information. Otherwise, the vehicle's instantaneous speed is estimated using a multi-target tracking algorithm based on the difference between the vehicle's first appearance position and its current position, as well as the vehicle's front direction information. Further, whether illegal parking has occurred is determined based on the vehicle's instantaneous speed, its positional relationship with the illegal parking area and the type of illegal parking.

[0021] Furthermore, it also includes:

[0022] Build a license plate detection model based on the YOLOv8 algorithm to detect the captured license plate image and obtain the key points and number of layers of the license plate;

[0023] Perform affine transformation on the detected key points of the license plate, segment the non-single-layer license plate and splice it into a single-layer license plate to obtain a processed license plate image;

[0024] A license plate number recognition model based on the CRNN algorithm is constructed to analyze the processed license plate image to obtain the license plate number.

[0025] Furthermore, the training data of the vehicle detection model and the license plate recognition model are obtained by the following method:

[0026] In addition to the original rotation target frame annotation information of the open source remote sensing dataset, the key point annotation information of the vehicle head direction is added, and the motor vehicle categories are unified;

[0027] Adjust the label information of the open source license plate recognition dataset to license plate key point location information, license plate number information, license plate color information and license plate layer number information;

[0028] Supplement the remote sensing data collected by the drone at different altitudes and pitch angles, and mark the rotating target frame, key points of the vehicle head direction and the type of motor vehicle;

[0029] Supplementary license plate recognition data is collected from drones at different altitudes and pitch angles, and the key points of the license plate, license plate number, license plate color and number of license plate layers are marked.

[0030] The present invention also provides a vehicle illegal parking control system based on a drone, which is used to implement the method described above, including a drone, a drone intelligent flight control module for controlling the drone to achieve automated cruising tasks, and a road illegal parking intelligent control module for identifying vehicle illegal parking incidents and controlling them based on cruising data collected by the drone.

[0031] Furthermore, the UAV intelligent flight control module includes:

[0032] A route management module is used to set the route management information of the drone; the route management information includes the flight status and flight segment, the flight status includes the flight altitude, flight speed, pitch angle, yaw angle, roll angle and pan / tilt camera focal length, and the flight segment includes the take-off point, each intermediate waypoint, the return point position and the corresponding flight status;

[0033] The task management module is used to set the cruise task and save the flight record according to the set cruise task and the illegal parking identification result; the cruise task setting includes filling in the name of the executor, loading the execution route, selecting the task type, and supporting the addition of a blacklist and whitelist for the executor. The task type includes optional immediate execution, scheduled execution, and periodic execution; the flight record includes the task start time, task batch information, executor name, illegal parking identification icon 1, icon 2, license plate number, location, illegal parking type, drone number, cloud box number, identification time, identification result, and review result;

[0034] The flight control management module is used to switch the flight mode according to various monitoring data monitored in real time to ensure the flight safety of the UAV; the monitoring data includes telemetry data and status data, the telemetry data includes the longitude and latitude, altitude, and speed of the UAV, and the status data includes battery power, satellite count, and obstacle distance; the flight mode includes take-off, return, route pause, and emergency braking.

[0035] Furthermore, the road illegal parking intelligent management and control module includes:

[0036] The illegal parking database is used to store road information, route management information and calibration information corresponding to each road information, parking space information, illegal parking areas and illegal parking types corresponding to different illegal parking areas; the road information includes road categories, which include highways, urban roads, special roads, and expressways; the parking space information includes parking space area and direction information; the illegal parking areas include fire lanes, sidewalks, non-motorized vehicle lanes, no parking sign control areas, no long-term parking sign control areas, specific time control areas, construction areas and emergency lane areas, in addition to railway crossings, narrow roads, bridges, steep slopes, tunnels and blind roads, as well as bus stops, emergency stations and fire stations where special vehicles are allowed to park;

[0037] The illegal parking identification module is used to determine whether illegal parking has occurred based on the cruise data collected by the drone and the parking space information, illegal parking area and illegal parking type in the illegal parking database;

[0038] The intelligent control module is used to link up with the police system for control after illegally parked motor vehicles are discovered.

[0039] Beneficial Effects

[0040] Due to the adoption of the above-mentioned technical scheme, the present invention has the following advantages and positive effects compared with the prior art: the present invention realizes automated cruising by presetting various parameters in the UAV cruising route; by arranging flight plans, it can orderly execute intelligent control tasks; by real-time monitoring of various data, switching flight modes to ensure flight safety; by establishing road information big data, providing data support for AI algorithms, so that UAVs can safely and efficiently execute illegal parking control tasks; through AI algorithms, decision support is provided for intelligent control, and illegal parking of motor vehicles can be quickly and accurately discovered; after discovering illegal parking of motor vehicles, the driver is promptly reminded to stop the illegal parking behavior through UAV payload shouting and SMS notification, and the alarm mechanism is triggered at the first time to facilitate traffic police control and prevent danger, ultimately improving the detection efficiency and accuracy of illegal parking of vehicles and realizing automated intelligent control. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a flow chart of a first embodiment of the present invention;

[0042] Figure 2 It is a system diagram of the second embodiment of the present invention. DETAILED DESCRIPTION

[0043] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0044] The first embodiment of the present invention relates to a method for controlling illegal parking of vehicles based on drones, such as Figure 1 As shown, the following steps are included:

[0045] Step 1: Based on the open source data set and combined with the characteristics of the system of the present invention, continue to expand new data to complete the training of rotating target detection and vehicle head orientation model, license plate positioning and recognition model. For example, based on the DroneVehicle and VisDrone remote sensing data sets, delete unnecessary categories, add key point annotation information of the vehicle head direction in addition to the original rotating target frame annotation information, unify the motor vehicle categories into cars, trucks, vans, buses, long-distance buses, police cars, fire trucks, ambulances, military vehicles, engineering vehicles, tank trucks, and supplement the data collected by drones at different heights and different pitch angles; based on the CCPD and CRPD license plate recognition data sets, adjust the label information to the key point position information of the license plate, license plate number information, license plate color information, and license plate layer information. Since the open source data set is taken by mobile phones, the angle and height are single, and there is a lack of new energy license plates, the drone continues to collect more license plate information through zooming at different heights and different pitch angles for supplementation.

[0046] Step 2: Based on the YOLOv8 algorithm and integrating the rotating target detection algorithm and the key point detection algorithm, the rotating vehicle detection and vehicle head orientation models are trained; based on the YOLOv8 key point detection algorithm and the CRNN algorithm, the license plate positioning and recognition model is trained, in which the detected license plate key points are affine transformed, and the double-layer license plate is segmented and spliced ​​into a single-layer license plate, which effectively improves the accuracy of license plate recognition.

[0047] Step 3: Initialize the illegal parking recognition AI algorithm and complete the loading of all basic models.

[0048] Step 4: Initialize illegal parking big data information. For example, if you monitor illegal parking of motor vehicles on a certain road section, the road category can be a general highway, etc. Set the corresponding route management information for the road section, such as setting the global flight speed in the flight section to 2m / s, the altitude to 50m, the load pitch angle to -60°, set the starting point and end point of the monitored section, and if there are two bends in the middle, you can set waypoint 1 and waypoint 2, adjust the heading of the drone at the waypoint, and keep the drone directly above the road surface; calibrate in advance under the route management information, the calibration scheme is such as on a 100m*100m flat ground, with the vertical take-off point as the vertical projection point of the drone, that is, the coordinate origin O, select the vertical upward as the Y axis, the horizontal right as the X axis, and the origin to the X-axis and Y-axis. The surveying and calibration are carried out at intervals of 10m. The UAV takes off to 50m, adjusts the payload pitch angle to -60°, hovers for data collection, performs distortion correction and affine transformation on the collected images, maps the coordinate relationship between the X-axis and the Y-axis based on the standard pixel coordinate system, and completes the conversion from pixel coordinates to geographic coordinates. There is no parking space information on this section of road. Otherwise, the vehicle position information in the world coordinates can be calculated based on the requirements of the vehicle head orientation and the parking space orientation, based on the UAV position coordinates, the vehicle's pixel coordinates and the corresponding ground position calibration, and the parking space area information is combined to determine whether parking is in compliance with regulations. The illegal parking area can be the entire road section, and the illegal parking type can be prohibition of long-term parking, which requires a first warning.

[0049] Step 5: In the task management, set the cruise task type to periodic execution, such as setting it from 9:00 to 18:00 from Monday to Friday, with a cruise interval of 20 minutes each time, and load the execution route.

[0050] Step 6: When the drone cruises in the route, it identifies ground vehicles that have been illegally parked. It detects all vehicle category information, rotating frame category information, and vehicle head orientation information through rotating vehicle detection and vehicle head orientation model; it comprehensively judges the vehicle's driving direction through vehicle tracking and vehicle head orientation information, the difference between the first appearance position and the current position, and defines the vehicle's instantaneous speed as the actual moving distance per second. Combined with the drone's flight speed, the weighted instantaneous speed of each vehicle is obtained. For example, if the drone's cruising speed is 8m / s, the actual speed of the vehicle in the same direction is the moving distance per second plus the drone's speed, and the actual speed of the vehicle in the opposite direction is the moving distance per second minus the drone's speed. Combine the drone's instantaneous speed, the relationship with the illegal parking area, and the current illegal parking type to comprehensively determine whether an illegal parking incident has occurred.

[0051] Step 7: After triggering the illegal parking of a motor vehicle, the first snapshot is taken, and the exit route command and gimbal adjustment position information are sent to the drone. The gimbal adjustment position is the midpoint of the rotating frame facing the end of the vehicle head. The gimbal will automatically turn to this position and implement zooming. After the adjustment is completed, the drone will notify the AI ​​end. The AI ​​end will locate and recognize the license plate of the enlarged image, and take a second snapshot, and send the first SMS or phone warning to the driver of the license plate number.

[0052] Step 8: For illegal parking areas that require a first warning, the drone resumes the route and continues cruising. If the vehicle is recognized again after 20 minutes, the evidence is uploaded and the police system is notified to facilitate traffic police control. For illegal parking areas that do not require a first warning, the evidence will be uploaded immediately on the route where the illegal parking incident was first discovered.

[0053] Step 9: If the end of the route has not been reached, continue on the current route and identify new parking violations. If the end of the route has been reached, return and wait for the next cruise cycle.

[0054] Step 10: If the periodic cruise is completed, end the cruise mission, otherwise continue to execute.

[0055] In order to implement the above method, the following intelligent control system for illegal parking of motor vehicles based on drones is constructed, including a drone intelligent flight control module and a road illegal parking intelligent control module.

[0056] 1. UAV intelligent flight control module, the main functions are:

[0057] (1) Route management: Automatic cruising is achieved by presetting various parameters in the drone's cruising route. This includes flight status settings and flight segment settings. Flight status settings refer to the flight altitude, speed, pitch angle, yaw angle, roll angle, gimbal camera focal length and other parameter settings; flight segment settings refer to the location of the take-off point, intermediate waypoints, and return point, as well as the corresponding flight status settings.

[0058] (2) Task management: By arranging flight plans, intelligent control tasks can be executed in an orderly manner. This includes setting cruise tasks and saving flight records. Setting cruise tasks means filling in the name of the executor, loading the execution route, and selecting the task type. The executor can add a blacklist or whitelist, and the task type can be selected for immediate execution, scheduled execution, and periodic execution. After the task is issued and executed, the flight record will be saved according to the task settings and the illegal parking identification results, including the task start time, task batch information, executor name, illegal parking identification icon 1, icon 2, license plate number, location, illegal parking type, drone number, cloud box number, identification time, identification result, and review result.

[0059] (3) Flight control management: Ensure flight safety by real-time monitoring of various data and switching flight modes. The monitored data include telemetry data such as the longitude and latitude, altitude, speed, battery power, satellite count, obstacle distance and other status data; flight modes include take-off, return, pause route, emergency braking and other modes.

[0060] 2. Intelligent road parking control module, the main functions are:

[0061] (1) Illegal parking big data: By establishing road information big data, data support is provided for AI algorithms, so that drones can safely and efficiently perform illegal parking control tasks. It includes road information, route management information, calibration information, parking space information, illegal parking areas, and illegal parking types. Among them, road information refers to road categories including highways, urban roads, special roads, and expressways; each road has the corresponding route management information mentioned above, so that drones can safely perform illegal parking monitoring tasks; calibration information refers to calibration information corresponding to different road categories, supporting the conversion of pixel coordinates and geographic coordinates; parking space information refers to parking space area and direction information; illegal parking areas refer to fire lanes, sidewalks, non-motorized vehicle lanes, no parking sign control areas, no long-term parking sign control areas, specific time control areas, construction areas, emergency lane areas, in addition to railway crossings, narrow roads, bridges, steep slopes, tunnels, blind roads, and bus stops, emergency stations, fire stations, etc. where special vehicles are allowed to park; illegal parking types refer to illegal parking of different types in different areas, whether it includes the first warning.

[0062] (2) Illegal parking identification algorithm: Through AI algorithms, it provides decision support for intelligent management and control, and can quickly and accurately detect illegal parking incidents of motor vehicles. Specifically, it refers to rotating vehicle detection and vehicle head direction detection based on the YOLOv8 algorithm and the integration of the rotating target detection algorithm and the key point detection algorithm; vehicle speed estimation based on the ByteTrack multi-target tracking algorithm; and positioning and identifying license plate numbers based on the YOLOv8 key point detection algorithm and the CRNN algorithm.

[0063] Intelligent control: After discovering illegal parking of motor vehicles, the drone payload will be used to shout and send SMS notifications to promptly remind the driver to stop the illegal parking behavior, and the alarm mechanism will be triggered immediately to facilitate traffic police control and prevent danger. The triggering of the alarm mechanism means that if illegal parking is found in areas where the first warning is not required, the shouting will be triggered immediately, and the evidence will be uploaded, and the police system will be linked for control; for areas that require the first warning, after the first warning shouting is invalid, the snapshot will be taken again after a period of time. If it is found again, the first warning will be deemed invalid, and the evidence will be uploaded together for both snapshots; each upload includes two illegal parking diagrams, one global diagram that completely contains the illegally parked vehicles, and one large diagram collected after adjusting the automatic zoom of the gimbal, which can clearly see the license plate number of the illegally parked vehicle.

[0064] The second embodiment of the present invention relates to a vehicle illegal parking control system based on a drone, which is used to implement the method in the first embodiment, such as Figure 2 As shown, it includes a drone, a drone intelligent flight control module for controlling the drone to achieve automated cruising tasks, and a road illegal parking intelligent control module for identifying and controlling vehicle illegal parking incidents based on the cruising data collected by the drone.

[0065] 1. UAV intelligent flight control module, including:

[0066] (1) Route management module: Automatic cruising is achieved by presetting various parameters in the drone's cruising route. This includes flight status settings and flight segment settings. Flight status settings refer to the flight altitude, speed, pitch angle, yaw angle, roll angle, gimbal camera focal length and other parameter settings; flight segment settings refer to the location of the take-off point, intermediate waypoints, return point and the corresponding flight status settings.

[0067] (2) Task management module: By arranging flight plans, intelligent control tasks can be executed in an orderly manner. This includes setting cruise tasks and saving flight records. Setting cruise tasks means filling in the name of the executor, loading the execution route, and selecting the task type. The executor can add a blacklist or whitelist, and the task type can be selected for immediate execution, scheduled execution, and periodic execution. After the task is issued and executed, the flight record will be saved according to the task settings and the illegal parking identification results, including the task start time, task batch information, executor name, illegal parking identification icon 1, icon 2, license plate number, location, illegal parking type, drone number, cloud box number, identification time, identification result, and review result.

[0068] (3) Flight control management module: ensures flight safety by real-time monitoring of various data and switching flight modes. The monitored data include telemetry data such as the longitude and latitude, altitude, speed, battery power, satellite count, obstacle distance and other status data; flight modes include take-off, return, pause route, emergency braking and other modes.

[0069] 2. Intelligent road parking control module, including:

[0070] (1) Illegal parking database: By establishing big data of road information, data support is provided for AI algorithms, so that drones can safely and efficiently perform illegal parking control tasks. It includes road information, route management information, calibration information, parking space information, illegal parking areas, and illegal parking types. Road information refers to road categories including highways, urban roads, special roads, and expressways; each road has the corresponding route management information mentioned above, so that drones can safely perform illegal parking monitoring tasks; calibration information refers to calibration information corresponding to different road categories, supporting the conversion of pixel coordinates and geographic coordinates; parking space information refers to parking space area and direction information; illegal parking areas refer to fire lanes, sidewalks, non-motorized vehicle lanes, no parking sign control areas, no long-term parking sign control areas, specific time control areas, construction areas, emergency lane areas, in addition to railway crossings, narrow roads, bridges, steep slopes, tunnels, blind roads, and bus stops, emergency stations, fire stations, etc. where special vehicles are allowed to park; illegal parking types refer to illegal parking of different types in different areas, whether it includes the first warning.

[0071] (2) Illegal parking identification module: Through AI algorithms, it provides decision support for intelligent management and control, and can quickly and accurately detect illegal parking incidents of motor vehicles. Specifically, it uses the YOLOv8 algorithm and integrates the rotating target detection algorithm and the key point detection algorithm to detect rotating vehicles and vehicle head direction; uses the ByteTrack multi-target tracking algorithm to estimate vehicle speed; and uses the YOLOv8 key point detection algorithm and the CRNN algorithm to locate and identify license plate numbers.

[0072] (3) Intelligent control module: After discovering illegal parking of a motor vehicle, the drone will use the payload to shout and send SMS notifications to promptly remind the driver to stop the illegal parking behavior, and trigger the alarm mechanism immediately to facilitate traffic police control and prevent danger. The triggering alarm mechanism means that if illegal parking is found in an area where no initial warning is required, the alarm will be triggered immediately, and the evidence will be uploaded, and the police system will be linked for control; for areas where the initial warning is required, after the initial warning is invalid, the camera will be taken again after a period of time. If it is found again, the first warning will be deemed invalid, and the evidence will be uploaded together for both snapshots; each upload includes two illegal parking diagrams, one global diagram that completely includes the illegally parked vehicle, and one large diagram collected after adjusting the automatic zoom of the gimbal, so that the license plate number of the illegally parked vehicle can be clearly seen.

Claims

1. A method for controlling illegal parking of vehicles based on drones, characterized in that: The following steps are involved: Obtain road information of the target controlled section; Arrange a flight plan of the drone based on the road information; Obtaining image information collected by the drone when cruising according to the flight plan; The image information is used to analyze whether ground vehicles on the target controlled road section have illegally parked. If illegal parking has occurred, the drone is ordered to take a snapshot, and then the drone is ordered to pause the current route and adjust the gimbal camera to take another snapshot to obtain a clear license plate image.

2. The method according to claim 1, characterized in that The step of arranging a flight plan of the UAV based on the road information includes: The route management information of the UAV is set based on the road information, wherein the route management information includes a flight state and a flight segment, wherein the flight state includes a flight altitude, a flight speed, a pitch angle, a yaw angle, a roll angle, and a focal length of a gimbal camera, and the flight segment includes a take-off point, each intermediate waypoint, a return point, and a corresponding flight state; The cruise mission of the drone is set, and the cruise mission includes a mission type, and the mission type includes immediate execution, scheduled execution, and periodic execution.

3. The method according to claim 2, characterized in that Before the step of acquiring the image information collected by the drone when cruising according to the flight plan, the method also includes the step of calibrating the drone under the set route management information.

4. The method according to claim 3, characterized in that The calibrating of the UAV under the set route management information includes: Establish a geographic coordinate system with the take-off point as the coordinate origin; From the coordinate origin, surveying and calibration are performed along the X-axis and Y-axis directions at set intervals to obtain the mapping relationship between pixel coordinates and geographic coordinates.

5. The method according to claim 1, characterized in that The analyzing, based on the image information, whether a ground vehicle on the target controlled road section has an illegal parking incident includes: Obtain parking space information, illegal parking areas, and illegal parking types corresponding to different illegal parking areas on the target controlled road section; Construct a vehicle detection model based on the YOLOv8 algorithm and integrating the rotating target detection algorithm and the key point detection algorithm to perform target recognition on the image information and obtain the rotating vehicle information and the vehicle head orientation information; If there is a parking area on the road section, whether illegal parking has occurred is determined based on the vehicle's front direction information, vehicle position information and parking space information. Otherwise, the vehicle's instantaneous speed is estimated using a multi-target tracking algorithm based on the difference between the vehicle's first appearance position and its current position, as well as the vehicle's front direction information. Further, whether illegal parking has occurred is determined based on the vehicle's instantaneous speed, its positional relationship with the illegal parking area and the type of illegal parking.

6. The method according to claim 5, characterized in that Also includes: Build a license plate detection model based on the YOLOv8 algorithm to detect the captured license plate image and obtain the key points and number of layers of the license plate; Perform affine transformation on the detected key points of the license plate, segment the non-single-layer license plate and splice it into a single-layer license plate to obtain a processed license plate image; A license plate number recognition model based on the CRNN algorithm is constructed to analyze the processed license plate image to obtain the license plate number.

7. The method according to claim 6, characterized in that The training data of the vehicle detection model and the license plate recognition model are obtained by the following method: In addition to the original rotation target frame annotation information of the open source remote sensing dataset, the key point annotation information of the vehicle head direction is added, and the motor vehicle categories are unified; Adjust the label information of the open source license plate recognition dataset to license plate key point location information, license plate number information, license plate color information and license plate layer number information; Supplement the remote sensing data collected by the drone at different altitudes and pitch angles, and mark the rotating target frame, key points of the vehicle head direction and the type of motor vehicle; Supplementary license plate recognition data is collected from drones at different altitudes and pitch angles, and the key points of the license plate, license plate number, license plate color and number of license plate layers are marked.

8. A vehicle illegal parking control system based on drones, characterized in that: Used to implement the method described in claims 1-7, including a drone, a drone intelligent flight control module for controlling the drone to achieve automated cruising tasks, and a road illegal parking intelligent control module for identifying and controlling vehicle illegal parking incidents based on cruising data collected by the drone.

9. The system according to claim 8, characterized in that The UAV intelligent flight control module includes: A route management module is used to set the route management information of the drone; the route management information includes the flight status and flight segment, the flight status includes the flight altitude, flight speed, pitch angle, yaw angle, roll angle and pan / tilt camera focal length, and the flight segment includes the take-off point, each intermediate waypoint, the return point position and the corresponding flight status; The task management module is used to set the cruise task and save the flight record according to the set cruise task and the illegal parking identification result; the cruise task setting includes filling in the name of the executor, loading the execution route, selecting the task type, and supporting the addition of a blacklist and whitelist for the executor. The task type includes optional immediate execution, scheduled execution, and periodic execution; the flight record includes the task start time, task batch information, executor name, illegal parking identification icon 1, icon 2, license plate number, location, illegal parking type, drone number, cloud box number, identification time, identification result, and review result; The flight control management module is used to switch the flight mode according to various monitoring data monitored in real time to ensure the flight safety of the UAV; the monitoring data includes telemetry data and status data, the telemetry data includes the longitude and latitude, altitude, and speed of the UAV, and the status data includes battery power, satellite count, and obstacle distance; the flight mode includes take-off, return, route pause, and emergency braking.

10. The system according to claim 8, characterized in that The road illegal parking intelligent management and control module includes: The illegal parking database is used to store road information, route management information and calibration information corresponding to each road information, parking space information, illegal parking areas and illegal parking types corresponding to different illegal parking areas; the road information includes road categories, which include highways, urban roads, special roads, and expressways; the parking space information includes parking space area and direction information; the illegal parking areas include fire lanes, sidewalks, non-motorized vehicle lanes, no parking sign control areas, no long-term parking sign control areas, specific time control areas, construction areas and emergency lane areas, in addition to railway crossings, narrow roads, bridges, steep slopes, tunnels and blind roads, as well as bus stops, emergency stations and fire stations where special vehicles are allowed to park; The illegal parking identification module is used to determine whether illegal parking has occurred based on the cruise data collected by the drone and the parking space information, illegal parking area and illegal parking type in the illegal parking database; The intelligent control module is used to link up with the police system for control after illegally parked motor vehicles are discovered.

Images