Traffic vehicle snapshot method and device based on unmanned aerial vehicle, equipment and medium

Abstract

The embodiment of the invention discloses a traffic vehicle snapshot method and device based on an unmanned aerial vehicle, equipment and a medium. A specific embodiment of the method comprises the following steps: determining an unmanned aerial vehicle inspection path based on a geographic information network system; according to the unmanned aerial vehicle inspection path, controlling an associated unmanned aerial vehicle to perform video acquisition on the no-parking area to obtain video stream data; performing target detection on the video stream data to obtain an illegal parking vehicle image set; real-time positioning data and camera operation parameters of the unmanned aerial vehicle are obtained, space coordinate conversion is carried out on the real-time positioning data and the camera operation parameters based on a space coordinate conversion method and the illegal parking vehicle image set, and a vehicle position data set is obtained; performing digital encryption on the illegal parking vehicle image set, the vehicle position data set and the system timestamp to obtain an illegal parking evidence file; and transmitting the illegal parking evidence file to a traffic law enforcement system, and triggering road network warning. According to the embodiment, the effective working time of the unmanned aerial vehicle can be prolonged, and the road network warning delay is reduced.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of computer technology, and more specifically to a method, apparatus, device, and medium for capturing images of traffic vehicles based on unmanned aerial vehicles (UAVs). Background Technology

[0002] With the increase in motor vehicles, the contradiction between parking demand and parking lot supply has gradually become apparent. Limited by insufficient parking resources, coupled with the weak safety awareness of some drivers, illegal parking is ubiquitous, leading to increasingly prominent traffic congestion and safety issues. Therefore, to regulate parking order and improve road capacity, it is necessary to capture illegally parked vehicles to ensure public transportation safety. Currently, there are three main methods for capturing illegally parked vehicles in traffic management: detecting illegal parking through fixed cameras installed along roadsides, capturing images of illegally parked vehicles using mobile enforcement vehicles, and capturing images of illegally parked vehicles using drones. Among these, drones, which estimate the approximate location of vehicles and capture images based on their own positioning data and camera angles, have received widespread attention for capturing illegally parked vehicles.

[0003] However, when using the above methods to capture illegally parked vehicles, the following technical problems often arise: By using fixed flight routes and ignoring actual road and environmental conditions, inspection efficiency is reduced. Furthermore, due to the limited positioning accuracy and battery life of drones, which are also susceptible to environmental interference, the spatial location judgment of illegally parked vehicles by drones is significantly inaccurate. This leads to frequent erroneous capture operations by drones, consuming excessive power, reducing the effective working time of drones, and delaying road network warnings.

[0004] The information disclosed in this background section is only intended to enhance the understanding of the background of the present disclosure concept, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure provide methods, apparatuses, electronic devices, and computer-readable media for capturing images of traffic vehicles based on unmanned aerial vehicles (UAVs) to address one or more of the technical problems mentioned in the background section above.

[0007] In a first aspect, some embodiments of this disclosure provide a method for capturing traffic vehicles based on unmanned aerial vehicles (UAVs). The method includes: determining a UAV inspection path based on a geographic information network system; controlling associated UAVs to collect video data of a no-parking zone according to the UAV inspection path, obtaining video stream data; performing target detection on the video stream data to obtain a set of illegally parked vehicle images; acquiring real-time positioning data and camera operation parameters of the UAV, and performing spatial coordinate transformation on the real-time positioning data and camera operation parameters based on a spatial coordinate transformation method and the illegally parked vehicle image set to obtain a vehicle location dataset; digitally encrypting the illegally parked vehicle image set, the vehicle location dataset, and a system timestamp to obtain an illegal parking evidence file; transmitting the illegal parking evidence file to a traffic enforcement system; and triggering a road network warning in response to the traffic enforcement system receiving the illegal parking evidence file.

[0008] Secondly, some embodiments of this disclosure provide a traffic vehicle capture device based on unmanned aerial vehicles (UAVs). The device includes: a determining unit configured to determine a UAV inspection path based on a geographic information network system; a control unit configured to control an associated UAV to capture video of a no-parking area according to the UAV inspection path, thereby obtaining video stream data; a detection unit configured to perform target detection on the video stream data, thereby obtaining a set of illegally parked vehicle images; a conversion unit configured to acquire real-time positioning data and camera operation parameters of the UAV, and to perform spatial coordinate conversion on the real-time positioning data and camera operation parameters based on a spatial coordinate conversion method and the set of illegally parked vehicle images, thereby obtaining a vehicle location dataset; an encryption unit configured to digitally encrypt the set of illegally parked vehicle images, the vehicle location dataset, and a system timestamp, thereby obtaining an illegal parking evidence file; and a transmission unit configured to transmit the illegal parking evidence file to a traffic enforcement system, and to trigger a road network warning in response to the traffic enforcement system receiving the illegal parking evidence file.

[0009] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0010] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0011] The above embodiments of this disclosure have the following beneficial effects: The drone-based traffic vehicle capture method of some embodiments of this disclosure can improve the effective working time of drones and reduce road network warning delays. Specifically, the reasons for the shortened effective working time of drones and the delay in road network warnings are: by using fixed flight routes and ignoring actual road and environmental conditions, the inspection efficiency is reduced. Furthermore, due to the limited positioning accuracy and battery life of drones, and their susceptibility to environmental interference, the spatial location judgment of illegally parked vehicles by drones has a large error, leading to frequent erroneous capture operations, excessive power consumption, and thus shortened effective working time and delayed road network warnings. Based on this, the drone-based traffic vehicle capture method of some embodiments of this disclosure firstly determines the drone inspection path based on a geographic information network system. This allows for the planning of a reasonable inspection route based on real-time environmental conditions, improving the capture efficiency of drones. Secondly, according to the aforementioned drone inspection path, the associated drones are controlled to collect video data of the no-parking area, obtaining video stream data. This allows for the acquisition of original video data of the no-parking area, while avoiding excessive image processing time due to image blur and multiple drone captures. Next, target detection is performed on the aforementioned video stream data to obtain a set of images of illegally parked vehicles. This yields high-confidence images of illegally parked vehicles. Then, the real-time positioning data and camera operating parameters of the aforementioned UAV are acquired. Based on a spatial coordinate transformation method and the aforementioned set of images of illegally parked vehicles, spatial coordinate transformation is performed on the real-time positioning data and camera operating parameters to obtain a vehicle location dataset. This avoids frequent zooming and repeated acquisition caused by inaccurate positioning. Then, the aforementioned set of images of illegally parked vehicles, the aforementioned vehicle location dataset, and the system timestamp are digitally encrypted to obtain a violation evidence file. This provides a tamper-proof and compliant evidence file. Finally, the aforementioned violation evidence file is transmitted to the traffic enforcement system, and in response to the traffic enforcement system receiving the violation evidence file, a road network warning is triggered. This enables real-time traffic alerts. Ultimately, this improves the effective working time of the UAV and reduces the delay in road network warnings. Attached Figure Description

[0012] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0013] Figure 1 This is a flowchart of some embodiments of the traffic vehicle capture method based on the present disclosure; Figure 2This is a schematic diagram of the structure of some embodiments of the traffic vehicle capture device based on the present disclosure; Figure 3 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0014] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0015] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0016] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0017] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0018] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0019] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] Figure 1 A flow 100 of some embodiments of a drone-based traffic vehicle capture method according to the present disclosure is shown. The drone-based traffic vehicle capture method includes the following steps: Step 101: Determine the UAV inspection path based on the geographic information network system.

[0021] In some embodiments, the execution entity (e.g., a server) of the above-described UAV-based traffic vehicle capture method can determine the UAV inspection path based on a geographic information network system.

[0022] In some optional implementations of certain embodiments, the aforementioned executing entity may determine the UAV inspection path based on a geographic information network system through the following steps: Step one: Based on the acquired capture task parameter set and the aforementioned Geographic Information Network (GIS), generate a spatial polygon set of no-parking areas and the corresponding road topology. In practice, the executing entity can generate no-parking road segment names by parsing the capture task parameter set. Then, the executing entity can query the road boundary information corresponding to the no-parking road segments from the GIS based on the no-parking road segment names. Next, the executing entity can use geofencing technology to convert the road boundary information into closed geometric shapes, obtaining the spatial polygon set of the no-parking areas. Finally, the executing entity can extract the connection relationships, directional attributes, and hierarchical structure between the no-parking road segments based on the GIS to generate the road topology corresponding to the no-parking areas. The road topology is used to characterize the connection logic between each no-parking road segment. The capture task parameter set can be no-parking road segment or area information pre-set by management personnel.

[0023] Step two: Based on the aforementioned spatial polygon set and road topology, construct a three-dimensional path mesh model. In practice, the executing entity can construct the three-dimensional path mesh model using a path connection algorithm. This path connection algorithm can include, but is not limited to, any of the following: Dijkstra's algorithm, A... Algorithms and the Floyd algorithm. The aforementioned 3D path mesh model can be a mesh-like 3D data structure composed of nodes (spatial points that the drone can occupy) and edges (paths that the drone can fly).

[0024] Step 3: Based on historical illegal parking data and real-time traffic flow data, perform spatiotemporal weight analysis on the aforementioned 3D path grid model to obtain an inspection priority weight set. In practice, the implementing entity can determine the initial illegal parking anomaly level for different road segments at different times (morning peak, evening peak, and nighttime) based on historical illegal parking data. Then, the implementing entity can correct the initial illegal parking anomaly level based on real-time traffic flow data to obtain the actual illegal parking anomaly level. For example, the initial illegal parking anomaly level corresponding to high traffic volume and low average speed (i.e., real-time traffic flow data shows traffic congestion) can be increased. The initial anomaly level corresponding to low traffic volume and high average speed (i.e., real-time traffic flow data shows no traffic congestion) can be decreased. Finally, the implementing entity can assign priority weights to the corresponding path grid units in the aforementioned 3D path grid model based on the aforementioned illegal parking anomaly level to obtain an inspection priority weight set. The aforementioned initial anomaly level can include a high anomaly level (characterizing a high frequency of illegal parking events), a medium anomaly level, and a low anomaly level. The aforementioned historical illegal parking data can be the frequency and temporal distribution of illegal parking incidents on various road sections over a past period (e.g., one month). The aforementioned real-time traffic flow data can be the current traffic volume and average vehicle speed on the road. The higher the weight of the aforementioned inspection priority, the greater the probability that the corresponding area needs priority inspection. As an example, the aforementioned enforcement entity can set the inspection priority weight of road sections with historically high rates of illegal parking and real-time traffic flow data showing traffic congestion to 0.9.

[0025] Step four: Based on the aforementioned inspection priority weight set and UAV endurance conditions, generate a preliminary inspection path. In practice, the executing entity can use a path optimization algorithm, with the aforementioned inspection priority weight set and UAV endurance conditions as constraints, to generate a preliminary inspection path. The path optimization algorithm can be a traveling salesman problem algorithm. The aforementioned UAV endurance conditions can be the UAV's maximum flight time and maximum flight distance. The aforementioned preliminary inspection path can include the UAV inspection path, flight coordinates (latitude and longitude), flight altitude, flight speed, and hovering point. The aforementioned UAV inspection path can include flight segments, flight start point, and flight end point.

[0026] Step 5: Based on real-time environmental data, perform anomaly assessment on the preliminary inspection path to obtain anomaly assessment information. In practice, the executing entity can determine the anomaly assessment information based on preset anomaly rules and real-time environmental data. The real-time environmental data may include wind speed, rainfall, air visibility, and temporary no-fly zones. The anomaly assessment information can be used to identify problematic sections (abnormal paths) in the drone inspection path, indicating whether the route is feasible or infeasible. For example, the anomaly assessment information could be that the route is infeasible because drone inspection path A (abnormal path A) passes through a strong wind area. The preset anomaly rules can be pre-defined rules for assessing drone inspection paths. For example, the anomaly assessment information could be that the route is infeasible in response to real-time wind speed exceeding the drone's safe flight threshold, or that the route is infeasible in response to the drone inspection path passing through a no-fly zone.

[0027] Step Six: In response to the feasibility of the above-mentioned anomaly assessment information characterization route, the above preliminary inspection path is determined as the UAV inspection path. The above-mentioned UAV inspection path includes: UAV inspection path, flight coordinates (latitude and longitude), flight altitude, flight speed, and hovering point.

[0028] Step 7: In response to the infeasibility of the above-mentioned anomaly assessment information representation route, perform the following path adjustment operations: Sub-step one: Based on the above anomaly assessment information, determine the abnormal paths of the preliminary inspection path to obtain an abnormal path area information set. This abnormal path area information set may include the abnormal path and its corresponding anomaly type. The anomaly types may include, but are not limited to, "meteorological anomaly: strong winds," "airspace anomaly: temporary no-fly zone," and "building obstacle anomaly: high-rise buildings."

[0029] Sub-step two involves selecting the corresponding path adjustment strategy from the path adjustment strategy library based on the aforementioned anomaly types. This path adjustment strategy library can be a predefined database of path adjustment schemes for various anomaly types. For example, for the "Weather Anomaly: Strong Winds" type, an strategy of increasing flight altitude can be adopted. For the "Airspace Anomaly: Temporary No-Fly Zone" type, a detour strategy can be adopted. For the "Building Obstacle Anomaly: High-Rise Buildings" type, a detour strategy can be adopted.

[0030] Sub-step three involves performing conflict detection on the aforementioned 3D path mesh model based on the path adjustment strategy described above, obtaining a candidate path set. As an example, the executing entity can respond to the path adjustment strategy's representation of "adopting a detour strategy" by marking the corresponding paths in the 3D path mesh model as impassable based on the abnormal paths. Then, the executing entity can generate a candidate path set using a path optimization algorithm, with the original flight start point, original flight end point, UAV endurance conditions, and the impassable paths in the 3D path mesh model as constraints. The executing entity can also respond to the path adjustment strategy's representation of "increasing flight altitude strategy." Sub-step four involves performing weight analysis on the aforementioned candidate path set to generate the optimal inspection path. In practice, the execution entity can traverse the aforementioned inspection priority weight set to determine the inspection priority weight of each candidate path in the candidate path set, and determine the candidate path with the highest priority weight as the optimal inspection path, thereby generating the optimal inspection path.

[0031] Sub-step five involves performing anomaly assessment on the optimal inspection path to obtain anomaly assessment information. Based on this anomaly assessment information indicating route feasibility, the optimal inspection path is then determined as the UAV inspection path. In practice, the executing entity can perform anomaly assessment on the optimal inspection path using the method described in "Step five" of step 101, obtain anomaly assessment information, and based on this anomaly assessment information indicating route feasibility, determine the optimal inspection path as the UAV inspection path.

[0032] Step 8: In response to the above abnormal assessment information indicating that the route is not feasible, perform the path adjustment operation again.

[0033] Step 102: Based on the drone inspection path, control the associated drones to collect video data of the no-parking area and obtain video stream data.

[0034] In some embodiments, the aforementioned execution entity may control associated drones to collect video data of no-parking areas based on the aforementioned drone inspection path, thereby obtaining video stream data.

[0035] In practice, the aforementioned implementing entity can control associated drones to collect video data of the no-parking area according to the aforementioned drone inspection path through the following steps: The first step is to generate a flight waypoint sequence and a hovering point information set based on the aforementioned UAV inspection path. The flight waypoint sequence can be an ordered set of coordinate points that the UAV will sequentially reach during its flight. The hovering point information set represents the location information where the UAV will capture images, and may include the coordinates of the points where the UAV needs to hover and the hovering duration.

[0036] The second step involves generating a flight control command set based on the aforementioned flight waypoint sequence and sending this command set to the UAV flight control system to control the UAV for flight and hovering. In practice, the executing entity can determine the specific attitude and control surface quantities required for UAV flight based on the PID control algorithm and the aforementioned flight waypoint sequence, and convert these specific attitude and control surface quantities into commands readable by the UAV flight control system, thus obtaining the flight control command set. Then, the executing entity can send the flight control command set to the UAV flight control system via a wireless data link (WLAN, cellular network, satellite communication) to control the UAV for flight and hovering.

[0037] The third step involves controlling the drone to enter a hovering state in response to the drone reaching any hovering point coordinate in the hovering point information set. In practice, the executing entity can obtain the drone's real-time coordinates through Real-time Kinematic (RTK) positioning. Then, in response to the difference between the real-time coordinates and the hovering point coordinates meeting a preset range, the drone can be controlled to enter a hovering state. This preset range can be the allowable difference between the drone's real-time coordinates and the hovering point coordinates, centered on the hovering point, and can include horizontal tolerance, vertical tolerance, and angular tolerance. The horizontal tolerance (X-axis, Y-axis) can be [-2 meters, 2 meters]. The vertical tolerance (Z-axis) can be [-0.5 meters, 0.5 meters]. The angular tolerance can be [-5°, 5°].

[0038] Fourth, in response to the drone entering hovering mode, control the drone to capture video data according to the preset frame rate and resolution, obtaining video stream data. For example, the preset frame rate could be 30 frames per second, and the preset resolution could be 1080p.

[0039] Step 103: Perform target detection on the video stream data to obtain a set of images of illegally parked vehicles.

[0040] In some embodiments, the aforementioned execution entity may perform target detection on the aforementioned video stream data to obtain a set of images of illegally parked vehicles.

[0041] In some optional implementations of certain embodiments, the aforementioned execution entity may perform target detection on the aforementioned video stream data through the following steps to obtain a set of images of illegally parked vehicles: Step one involves encoding the spatiotemporal features of the aforementioned video stream data to construct a spatiotemporal feature pyramid set. In practice, the executing entity can encode the video stream data using 3D-Convolutional Neural Networks (3D-CNN) to extract spatial information (e.g., vehicle appearance) and temporal information (e.g., vehicle motion) features contained within the video stream data. Then, the executing entity can use a Feature Pyramid Network (FPN) to extract features from the spatial and temporal information features, obtaining the spatiotemporal feature pyramid set.

[0042] Step 2: Perform target detection on the above spatiotemporal feature pyramid set to obtain an initial set of vehicle detection bounding box images.

[0043] In addressing the technical challenges of vehicle detection in drone video streams, the following issues arise in the application scenario: When drones perform high-altitude inspections, they need to acquire clear vehicle images. However, this often presents several problems: visually similar interfering objects in complex urban environments (such as shadows and boxes), environmental factors (such as rain, fog, and glare), and image jitter caused by the drone's own flight. Relying solely on a single vehicle recognition result to control the drone's image acquisition can easily lead to multiple invalid shots. Furthermore, the drone's limited load capacity results in wasted memory resources and low computational resource utilization. Considering the following requirements for this application scenario: interference resistance and low false detection rate, we have decided to adopt the following solution: In practice, the aforementioned executing entity can perform target detection on the aforementioned spatiotemporal feature pyramid set through the following steps to obtain an initial set of vehicle detection bounding box images: Sub-step one involves determining the spatial features of the aforementioned spatiotemporal feature pyramid set to generate a spatial feature weight set. In practice, the executing entity can use the L2 norm to determine the feature activation intensity of each spatiotemporal feature pyramid in the aforementioned spatiotemporal feature pyramid set, obtaining a feature activation intensity set. Then, the executing entity can assign weights to each feature activation intensity in the aforementioned feature activation intensity set, obtaining a spatial feature weight set. Specifically, the executing entity can determine the sum of all feature activation intensities in the aforementioned feature activation intensity set, and then divide each feature activation intensity by the sum to obtain the spatial feature weight set.

[0044] Sub-step two involves pruning the spatiotemporal feature pyramid set based on the aforementioned spatial feature weight set to obtain a sparse feature pyramid set. In practice, the executing entity can obtain the sparse feature pyramid set by removing the spatiotemporal feature pyramids corresponding to spatial feature weights below a preset weight threshold from the aforementioned spatial feature weight set. The preset weight threshold can be 0.6.

[0045] Sub-step three involves enhancing the sparse feature pyramid set based on environmental data to obtain an anti-interference feature set. In practice, the executing entity can invoke the corresponding feature enhancement algorithm based on the weather conditions represented by the environmental data to enhance the sparse feature pyramid set and obtain an anti-interference feature set. The environmental data can be real-time weather information (such as fog, rain, clear skies, and snow). The feature enhancement algorithms can include, but are not limited to, de-raining algorithms and de-fogging algorithms. For example, in response to the environmental data indicating foggy weather, the executing entity can use a de-fogging algorithm to enhance the sparse feature pyramid set and obtain an anti-interference feature set. The de-fogging algorithm can include, but is not limited to, any of the following: the ACE de-fogging algorithm and a de-fogging algorithm based on Dark Channel Prior (DCP). Similarly, in response to the environmental data indicating rainy weather, the executing entity can use a de-raining algorithm to enhance the sparse feature pyramid set and obtain an anti-interference feature set. The de-raining algorithm can be the NeRD-Rain algorithm.

[0046] Sub-step four involves performing background compensation on the aforementioned anti-interference feature set based on the acquired real-time UAV pose data, resulting in a compensated feature set. In practice, the executing entity can determine the UAV movement vector between consecutive frames based on the UAV's real-time pose data. Then, the executing entity can perform a reverse translation on the aforementioned anti-interference feature set based on the UAV movement vector to offset the background shift, thereby obtaining the compensated feature set.

[0047] Sub-step five involves performing cross-attention matching on the aforementioned compensated feature set and the pre-trained vehicle feature set to generate a moving vehicle feature set. In practice, the aforementioned execution entity can use a cross-attention mechanism to determine the similarity between the compensated feature set and the pre-trained vehicle feature set, obtaining the vehicle features with the highest similarity to the compensated feature set to generate the moving vehicle feature set. The aforementioned pre-trained vehicle feature set can be a dataset containing various typical vehicle features. These typical vehicle features can include geometric appearance features (such as sedans, trucks, vans, etc.), component features (such as wheel styles, window features, headlight features, etc.), and material features (such as vehicle color).

[0048] Sub-step six involves performing network forward propagation on the aforementioned moving vehicle feature set to generate a candidate vehicle region set and a corresponding foreground score set. In practice, the aforementioned execution entity can use a Region Proposal Network (RPN) to perform network forward propagation on the aforementioned moving vehicle feature set to generate a candidate vehicle region set and a corresponding foreground score set. The candidate vehicle regions in the aforementioned candidate vehicle region set can be rectangular regions containing vehicles. The foreground scores in the aforementioned foreground score set can be used to characterize the probability that the region contains a vehicle.

[0049] Sub-step seven involves filtering the candidate vehicle region set based on the aforementioned foreground score set to obtain an initial vehicle set. In practice, the executing entity can remove candidate vehicle regions corresponding to foreground scores below a preset value (e.g., 0.7) from the aforementioned foreground score set to obtain a high-scoring vehicle region set. Then, the executing entity can use the Non-Maximum Suppression (NMS) algorithm to remove highly overlapping high-scoring vehicle regions to obtain the initial vehicle set.

[0050] Sub-step eight: Based on the aforementioned initial vehicle set, control the UAV to adjust its flight elevation angle to obtain an initial vehicle detection box image set. The initial vehicle detection box images in this set can be images where the target vehicle is centered in the frame and marked with a rectangular frame.

[0051] The aforementioned sub-steps one through eight and their related content, as an inventive point of this disclosure, solve the technical problem that "due to visually similar interference objects (such as shadows, boxes), environmental factors (such as rain, fog, shadow reflections), and image jitter caused by the drone's own flight in complex urban environments, relying solely on a single vehicle recognition result to control the drone to acquire vehicle images easily leads to multiple invalid shots by the drone, and the drone's own limited load capacity easily wastes memory resources, resulting in low computing resource utilization." The reason for low computing resource utilization is that, due to visually similar interference objects (such as shadows, boxes), environmental factors (such as rain, fog, shadow reflections), and image jitter caused by the drone's own flight in complex urban environments, relying solely on a single vehicle recognition result to control the drone to acquire vehicle images easily leads to multiple invalid shots by the drone, and the drone's own limited load capacity easily wastes memory resources, resulting in low computing resource utilization. Solving these factors can resolve the problem of low computing resource utilization. To achieve this effect, the first step is to determine the spatial features of the aforementioned spatiotemporal feature pyramid set to generate a spatial feature weight set. This allows for data filtering, ensuring that only high-weight features are processed subsequently, thus reducing computational load. The second step involves pruning the spatiotemporal feature pyramid set based on the aforementioned spatial feature weight set, resulting in a sparse feature pyramid set. This reduces data volume by removing redundant low-weight features. The third step involves feature enhancement on the sparse feature pyramid set based on environmental data, resulting in an anti-interference feature set. This reduces recognition errors caused by environmental factors. The fourth step involves background compensation on the anti-interference feature set based on the acquired real-time UAV pose data, resulting in a compensated feature set. This reduces recognition errors caused by the UAV's own motion. The fifth step involves cross-attention matching between the compensated feature set and the pre-trained vehicle feature set to generate a moving vehicle feature set. This reduces recognition errors caused by visual interference (such as shadows and boxes). The sixth step involves forward propagation of the moving vehicle feature set to generate a candidate vehicle region set and a corresponding foreground score set. This yields high-quality candidate vehicles. The seventh step involves filtering the candidate vehicle region set based on the foreground score set to obtain an initial vehicle set. Therefore, threshold filtering can be used to obtain more reliable vehicle information. The eighth step involves adjusting the drone's flight elevation angle based on the initial vehicle set to acquire an initial set of vehicle detection bounding box images. This ensures that each drone shot acquires valid vehicle images, avoiding unnecessary consumption of drone battery power and memory resources. Ultimately, this solves the problem of low computational resource utilization.

[0052] Step 3: Perform cross-frame association on the initial vehicle detection box image set to obtain a vehicle trajectory fragment set. In practice, the aforementioned execution entity can use a multi-object tracking (MOT) algorithm to connect the same vehicle appearing in different frames to generate vehicle trajectory fragments, thus obtaining a vehicle trajectory fragment set.

[0053] Step four involves performing motion state analysis on the aforementioned vehicle trajectory fragment set to obtain a candidate set of stationary vehicle trajectories. In practice, the executing entity can determine the speed of each vehicle in the candidate set of vehicle trajectories, and in response to speeds below a preset speed threshold (e.g., 5 km / h), determine the corresponding vehicle trajectory as a stationary vehicle trajectory, thus obtaining a candidate set of stationary vehicle trajectories.

[0054] Step 5: Based on the aforementioned geographic information network system, perform spatial overlay analysis on the candidate set of stationary vehicle trajectories to obtain the set of vehicle trajectories for the no-parking area. In practice, the executing entity can determine whether the coordinates of each stationary vehicle trajectory in the candidate set of stationary vehicle trajectories are located within the no-parking area based on the aforementioned geographic information network system and the point and polygon testing algorithm. Then, in response to the fact that the coordinates of the stationary vehicle trajectory are located within the no-parking area, the executing entity can identify the stationary vehicle trajectory as a vehicle trajectory for the no-parking area, thus obtaining the set of vehicle trajectories for the no-parking area.

[0055] Step Six: Based on the preset dwell time and the aforementioned set of vehicle trajectories in the no-parking zone, determine the set of illegally parked vehicle trajectories. In practice, the executing entity can determine the no-parking zone vehicle trajectory corresponding to the aforementioned vehicle as an illegally parked vehicle trajectory if the dwell time of the vehicle in the aforementioned set of vehicle trajectories in the no-parking zone exceeds the preset dwell time, thus obtaining the set of illegally parked vehicle trajectories. The preset dwell time can be 30 seconds.

[0056] Step seven involves extracting keyframes from the aforementioned set of illegally parked vehicle trajectories to obtain a set of illegally parked vehicle images. In practice, the executing entity can select video frame images that meet preset image standards from the video stream data corresponding to each illegally parked vehicle trajectory in the aforementioned set of illegally parked vehicle trajectories to obtain the set of illegally parked vehicle images.

[0057] Step 104: Obtain the real-time positioning data and camera operation parameters of the UAV, and perform spatial coordinate transformation on the real-time positioning data and camera operation parameters based on the spatial coordinate transformation method and the image set of illegally parked vehicles to obtain the vehicle location dataset.

[0058] In some embodiments, the aforementioned execution entity may acquire the real-time positioning data and camera operation parameters of the aforementioned UAV, and based on the spatial coordinate transformation method and the aforementioned illegally parked vehicle image set, perform spatial coordinate transformation on the aforementioned real-time positioning data and camera operation parameters to obtain a vehicle location dataset.

[0059] In addressing the technical challenges of vehicle detection via drone video streams, the following issues arise when using drones to locate illegally parked vehicles in complex urban environments: drone positioning and attitude errors, light pollution in urban environments, and significant errors (up to tens of meters) in determining vehicle coordinates using traditional GPS systems lead to inaccurate positioning, causing the drone to reach incorrect locations. This results in excessive battery consumption due to continuous camera zooming to capture images of illegally parked vehicles, ultimately shortening the drone's effective operating time. Considering the specific requirements of this application scenario—accurate positioning and limited battery life—we have decided to adopt the following solution: In some optional implementations of certain embodiments, the aforementioned executing entity may obtain the real-time positioning data and camera operation parameters of the aforementioned UAV through the following steps, and perform spatial coordinate transformation on the aforementioned real-time positioning data and camera operation parameters based on the spatial coordinate transformation method and the aforementioned illegally parked vehicle image set to obtain a vehicle location dataset: Step one: Acquire the real-time positioning data, attitude data, and camera operating parameters of the configured UAV. In practice, the executing entity can acquire the UAV's real-time positioning data through real-time dynamic positioning. Attitude data can be acquired through an Inertial Measurement Unit (IMU). The camera operating parameters include focal length, gimbal angle, and lens distortion parameters. The attitude data can include pitch angle, elevation angle, and yaw angle.

[0060] Step two: Based on the aforementioned camera operating parameters, construct a camera imaging model. This camera imaging model can be a matrix model containing parameters such as focal length and optical coordinates, used to convert three-dimensional point coordinates into pixel coordinates on the image.

[0061] Step 3: Perform tight-coupled decomposition on the aforementioned real-time positioning data and attitude data to obtain the UAV pose set. In practice, the aforementioned execution entity can use the Kalman filter algorithm to perform tight-coupled decomposition on the aforementioned real-time positioning data and attitude data to obtain the UAV pose set.

[0062] Step four: Based on the aforementioned camera imaging model and the aforementioned set of images of illegally parked vehicles, determine the vehicle orientation vector set. In practice, the aforementioned execution entity can use the aforementioned camera imaging model to back-project the aforementioned set of images of illegally parked vehicles to obtain the vehicle orientation vector set. This vehicle orientation vector set is used to characterize the vehicle's orientation relative to the UAV.

[0063] Step 5: Based on the station-centered coordinate system and the aforementioned vehicle direction vector set, determine the vehicle ray vector set. In practice, the aforementioned executing entity can use a station-centered coordinate system (ENU) with the current position of the UAV as the origin to perform coordinate transformation on each vehicle direction vector in the aforementioned vehicle direction vector set to obtain the vehicle ray vector set.

[0064] Step six: Control the drone to measure the distance to illegally parked vehicles to generate a vehicle slant distance dataset. The vehicle slant distance data in this dataset can be the straight-line distance between the drone and the vehicle.

[0065] Step 7: Using the spatial forward intersection algorithm, solve the aforementioned vehicle slant range dataset, UAV pose set, and vehicle ray vector set to obtain the vehicle position coordinate set. In practice, the executing entity can use the spatial forward intersection algorithm to determine the three-dimensional points of the aforementioned vehicle slant range dataset and vehicle ray vector set in the station-centered coordinate system, obtaining a three-dimensional point set. Then, based on the aforementioned UAV pose set, perform coordinate transformation on the aforementioned three-dimensional point set to obtain the vehicle position coordinate set. The aforementioned vehicle position coordinate set can represent the vehicle's latitude, longitude, and altitude coordinates in the geodetic coordinate system.

[0066] Step eight: Based on the aforementioned geographic information network system, the vehicle location coordinate set is encrypted and encapsulated to obtain a vehicle location dataset. In practice, the executing entity can use an encryption algorithm to encrypt and encapsulate the vehicle location coordinate set. The vehicle location dataset includes vehicle location coordinates and the name of the road segment to which they belong. The encryption algorithm can be the AES-256 algorithm.

[0067] Step nine: Based on the aforementioned vehicle location dataset, control the drone to descend and adjust the camera focus to obtain an initial set of vehicle images. In practice, the executing entity can obtain the initial set of vehicle images by controlling the drone to descend and adjust the camera focus, or it can use the vehicle location data in the aforementioned vehicle location dataset to call the corresponding monitoring facilities to obtain the initial set of vehicle images. The initial vehicle images in the aforementioned initial set of vehicle images can include vehicle images including license plate numbers.

[0068] The aforementioned sub-steps one through eight and their related content constitute an inventive point of this disclosure, solving the technical problem that "due to positioning and attitude errors of the drone, as well as light pollution in the urban environment, the determination of vehicle coordinates through the traditional GPS positioning system can result in significant errors (up to tens of meters), causing the drone to arrive at the wrong location due to inaccurate positioning. This leads to continuous camera zooming to photograph illegally parked vehicles, wasting excessive power and consequently shortening the drone's effective working time." The reason for this shortened effective working time is that the drone's positioning and attitude errors, as well as light pollution in the urban environment, and the significant errors (up to tens of meters) in determining vehicle coordinates through the traditional GPS positioning system, cause the drone to arrive at the wrong location due to inaccurate positioning. This leads to continuous camera zooming to photograph illegally parked vehicles, wasting excessive power and consequently shortening the drone's effective working time. Solving these factors can resolve the problem of shortened effective working time for drones. To achieve this, the first step is to acquire the drone's real-time positioning data, attitude data, and camera operating parameters of the configured camera. This yields the relevant raw data of the drone. The second step is to construct a camera imaging model based on the aforementioned camera operating parameters. This provides the correct geometric relationship for vehicle positioning. The third step involves tightly coupling the aforementioned real-time positioning data and attitude data to obtain the UAV pose set. This avoids and eliminates the influence of UAV attitude on positioning. The fourth step, based on the aforementioned camera imaging model and the aforementioned set of images of illegally parked vehicles, determines the vehicle orientation vector set. This determines the vehicle's orientation relative to the UAV. The fifth step, based on the stationary coordinate system and the aforementioned vehicle orientation vector set, determines the vehicle ray vector set. This allows for a direct association between coordinates and geographic space. The sixth step involves controlling the UAV to measure the distance to the illegally parked vehicles to generate a vehicle slant distance dataset. This avoids errors inherent in traditional visual positioning. The seventh step, using a spatial forward intersection algorithm, solves the aforementioned vehicle slant distance dataset, the aforementioned UAV pose set, and the aforementioned vehicle ray vector set to obtain the vehicle position coordinate set. This provides the location information of the illegally parked vehicles. The eighth step, based on the aforementioned geographic information network system, encrypts and encapsulates the aforementioned vehicle position coordinate set to obtain a vehicle position dataset. This enables encrypted storage of vehicle information, ensuring that the information is not tampered with, thus meeting law enforcement needs. The ninth step involves controlling the drone to descend and adjust the camera focus based on the aforementioned vehicle location dataset to acquire an initial set of vehicle images. This allows the drone to capture images based on vehicle location, ultimately increasing the drone's effective operating time.

[0069] Step 105: Digitally encrypt the image set of illegally parked vehicles, the vehicle location dataset, and the system timestamp to obtain the evidence file of illegal parking.

[0070] In some embodiments, the executing entity may digitally encrypt the above-mentioned illegally parked vehicle image set, the above-mentioned vehicle location dataset, and the system timestamp to obtain an evidence file of illegal parking.

[0071] In some optional implementations of certain embodiments, the executing entity may digitally encrypt the illegally parked vehicle image set, the vehicle location dataset, and the system timestamp using the following steps to obtain an evidence file of illegal parking: Step one: Based on the preset evidence collection standards, the aforementioned vehicle location dataset, and the system timestamp, generate evidence description information. The preset evidence collection standards may be the "Technical Specification for Image Evidence Collection of Road Traffic Safety Violations" (GA / T832). The system timestamp may be generated by the drone system and is used to characterize the time when the drone captured the information of the illegally parked vehicle. As an example, the system timestamp could be 2025-01-01 12:30:05.

[0072] Step two involves converting the aforementioned evidence description information into a binary encoded stream. In practice, the executing entity can use an encoding protocol to convert the evidence description information into a binary encoded stream composed of binary numbers (0s and 1s). This encoding protocol can be ASCII or UTF-8 encoding.

[0073] Step three involves encrypting the aforementioned binary encoded stream with a digital watermark to generate a digital watermark signal set. In practice, the executing entity can use low-frequency digital embedding technology to convert the binary encoded stream into a digital watermark signal, thus obtaining the digital watermark signal set.

[0074] Step four: Embed the aforementioned digital watermark signal set into the aforementioned illegally parked vehicle image set to obtain the illegally parked image set. In practice, the executing entity can use Discrete Cosine Transform (DCT) technology to embed the aforementioned digital watermark signal set into the aforementioned illegally parked vehicle image set to obtain the illegally parked image set.

[0075] Step five involves associating and encapsulating the aforementioned illegal parking image set and the aforementioned evidence description information to obtain the illegal parking evidence file. In practice, the executing entity can use the traffic police's off-site enforcement system's receiving standards as a basis to associating and encapsulate the aforementioned illegal parking image set, the aforementioned evidence description information, and the aforementioned initial vehicle image to obtain the illegal parking evidence file.

[0076] Step 106: Transmit the evidence of illegal parking to the traffic enforcement system, and in response to the traffic enforcement system receiving the evidence of illegal parking, trigger a road network warning.

[0077] In some embodiments, the aforementioned enforcement entity may transmit the aforementioned evidence of illegal parking to the traffic enforcement system, and in response to the traffic enforcement system receiving the aforementioned evidence of illegal parking, trigger a road network warning.

[0078] In practice, the aforementioned enforcement entities can transmit the illegal parking evidence documents to the traffic enforcement system via network protocols to generate traffic tickets. Then, based on the evidence documents, the enforcement entities can generate warning information and send it to traffic guidance facilities on the road where the illegally parked vehicle is located. Finally, based on the illegal parking evidence documents, the enforcement entities can obtain the vehicle owner's information (such as name and contact information) through the traffic management information platform to generate departure information and send it to the vehicle owner to remind them to leave the no-parking zone, thus completing the road network warning. As an example, the enforcement entities can send the departure information to the vehicle owner via SMS or a dedicated traffic management program to complete the road network warning. The network protocol can be an HTTPS protocol based on a national cryptographic algorithm (such as SM4). The warning information can be used to remind nearby drivers to pay attention to illegally parked vehicles and may include information about the illegal parking area and suggested detour routes. The traffic guidance facilities may include traffic broadcast systems and navigation map service providers (such as Gaode Maps and Baidu Maps).

[0079] The above embodiments of this disclosure have the following beneficial effects: The drone-based traffic vehicle capture method of some embodiments of this disclosure can improve the effective working time of drones and reduce road network warning delays. Specifically, the reasons for the shortened effective working time of drones and the delay in road network warnings are: by using fixed flight routes and ignoring actual road and environmental conditions, the inspection efficiency is reduced. Furthermore, due to the limited positioning accuracy and battery life of drones, and their susceptibility to environmental interference, the spatial location judgment of illegally parked vehicles by drones has a large error, leading to frequent erroneous capture operations, excessive power consumption, and thus shortened effective working time and delayed road network warnings. Based on this, the drone-based traffic vehicle capture method of some embodiments of this disclosure firstly determines the drone inspection path based on a geographic information network system. This allows for the planning of a reasonable inspection route based on real-time environmental conditions, improving the capture efficiency of drones. Secondly, according to the aforementioned drone inspection path, the associated drones are controlled to collect video data of the no-parking area, obtaining video stream data. This allows for the acquisition of original video data of the no-parking area, while avoiding excessive image processing time due to image blur and multiple drone captures. Next, target detection is performed on the aforementioned video stream data to obtain a set of images of illegally parked vehicles. This yields high-confidence images of illegally parked vehicles. Then, the real-time positioning data and camera operating parameters of the aforementioned UAV are acquired. Based on a spatial coordinate transformation method and the aforementioned set of images of illegally parked vehicles, spatial coordinate transformation is performed on the real-time positioning data and camera operating parameters to obtain a vehicle location dataset. This avoids frequent zooming and repeated acquisition caused by inaccurate positioning. Then, the aforementioned set of images of illegally parked vehicles, the aforementioned vehicle location dataset, and the system timestamp are digitally encrypted to obtain a violation evidence file. This provides a tamper-proof and compliant evidence file. Finally, the aforementioned violation evidence file is transmitted to the traffic enforcement system, and in response to the traffic enforcement system receiving the violation evidence file, a road network warning is triggered. This enables real-time traffic alerts. Ultimately, this improves the effective working time of the UAV and reduces the delay in road network warnings.

[0080] Further reference Figure 2 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a traffic vehicle capture device based on unmanned aerial vehicles (UAVs). These device embodiments are similar to... Figure 2 Corresponding to the method embodiments shown, the device can be specifically applied to various electronic devices.

[0081] like Figure 2As shown, some embodiments of the UAV-based traffic vehicle capture device 200 include: a determination unit 201, a control unit 202, a detection unit 203, a conversion unit 204, an encryption unit 205, and a transmission unit 206. The system comprises the following components: a determination unit 201, configured to determine the UAV inspection path based on a geographic information network system; a control unit 202, configured to control associated UAVs to collect video data of the no-parking area according to the UAV inspection path; a detection unit 203, configured to perform target detection on the video stream data to obtain a set of illegally parked vehicle images; a conversion unit 204, configured to acquire the real-time positioning data and camera operation parameters of the UAV, and to perform spatial coordinate conversion on the real-time positioning data and camera operation parameters based on a spatial coordinate conversion method and the set of illegally parked vehicle images to obtain a vehicle location dataset; an encryption unit 205, configured to digitally encrypt the set of illegally parked vehicle images, the vehicle location dataset, and the system timestamp to obtain an illegal parking evidence file; and a transmission unit 206, configured to transmit the illegal parking evidence file to the traffic enforcement system and, in response to the traffic enforcement system receiving the illegal parking evidence file, trigger a road network warning.

[0082] It is understandable that the units described in the device 200 are related to the reference. Figure 2 The steps in the described method correspond to each other. Therefore, the operations, features, and beneficial effects described above for the method also apply to the device 200 and the units contained therein, and will not be repeated here.

[0083] The following is for reference. Figure 3 It shows a schematic diagram of the structure of an electronic device (e.g., an electronic device) 300 suitable for implementing some embodiments of the present disclosure. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0084] like Figure 3 As shown, the electronic device 300 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. The RAM 303 also stores various programs and data required for the operation of the electronic device 300. The processing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0085] Typically, the following devices can be connected to I / O interface 305: input devices 306 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 307 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 308 including, for example, magnetic tapes, hard disks, etc.; and communication devices 309. Communication device 309 allows electronic device 300 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 An electronic device 300 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 3 Each box shown can represent a device or multiple devices as needed.

[0086] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 309, or installed from storage device 308, or installed from ROM 302. When the computer program is executed by processing device 301, it performs the functions defined in the methods of some embodiments of this disclosure.

[0087] It should be noted that, in some embodiments of this disclosure, the computer-readable medium described above may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0088] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0089] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: determine a drone inspection path based on a geographic information network system; control associated drones to collect video data of no-parking areas according to the drone inspection path, obtaining video stream data; perform target detection on the video stream data, obtaining a set of images of illegally parked vehicles; acquire the real-time positioning data and camera operation parameters of the aforementioned drones, and perform spatial coordinate transformation on the real-time positioning data and camera operation parameters based on a spatial coordinate transformation method and the set of images of illegally parked vehicles, obtaining a vehicle location dataset; digitally encrypt the set of images of illegally parked vehicles, the vehicle location dataset, and a system timestamp to obtain a violation evidence file; transmit the violation evidence file to the traffic enforcement system, and trigger a road network warning in response to the traffic enforcement system receiving the violation evidence file.

[0090] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0091] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0092] The units described in some embodiments of this disclosure can be implemented in software or hardware. The described units can also be housed in a processor; for example, a processor may be described as including a determining unit, a controlling unit, a detecting unit, a converting unit, an encryption unit, and a transmitting unit. The names of these units do not necessarily limit the specific unit; for example, the determining unit may also be described as "a unit that determines the inspection path of a UAV based on a geographic information network system."

[0093] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0094] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for capturing images of traffic vehicles based on unmanned aerial vehicles (UAVs), comprising: Determine the inspection path for drones based on a geographic information network system; Based on the drone inspection path, control the associated drones to collect video data of the no-parking area and obtain video stream data; Target detection is performed on the video stream data to obtain a set of images of illegally parked vehicles; The real-time positioning data and camera operation parameters of the UAV are obtained, and the spatial coordinate transformation is performed on the real-time positioning data and camera operation parameters based on the spatial coordinate transformation method and the image set of illegally parked vehicles to obtain the vehicle location dataset. The illegally parked vehicle image set, the vehicle location dataset, and the system timestamp are digitally encrypted to obtain the illegal parking evidence file; The evidence of illegal parking is transmitted to the traffic enforcement system, and in response to the traffic enforcement system receiving the evidence of illegal parking, a road network warning is triggered.

2. The method according to claim 1, wherein, The determination of the UAV inspection path based on the geographic information network system includes: Based on the acquired set of capture task parameters and the geographic information network system, a set of spatial polygons for the no-parking zone and the corresponding road topology are generated. Based on the spatial polygon set and the road topology, a three-dimensional path mesh model is constructed; Based on historical illegal parking data and real-time traffic flow data, spatiotemporal weight analysis is performed on the three-dimensional path grid model to obtain the inspection priority weight set. Based on the inspection priority weight set and the UAV's endurance conditions, a preliminary inspection path is generated; Based on real-time environmental data, an anomaly assessment is performed on the preliminary inspection path to obtain anomaly assessment information; In response to the feasibility of the anomaly assessment information characterization route, the preliminary inspection path is determined as the UAV inspection path, wherein the UAV inspection path includes: flight coordinates, flight altitude, flight speed, and hovering point.

3. The method according to claim 2, wherein, The method further includes: In response to the infeasibility of the aforementioned anomaly assessment information representation route, the following path adjustment operation is performed: Based on the anomaly assessment information, the abnormal path of the preliminary inspection path is determined, and an abnormal path area information set is obtained, wherein the abnormal path area information set includes the abnormal path area and the corresponding anomaly type. Based on the aforementioned anomaly type, select the corresponding path adjustment strategy from the path adjustment strategy library; Based on the path adjustment strategy, conflict detection is performed on the three-dimensional path mesh model to obtain a candidate path set; Weight analysis is performed on the candidate path set to generate the optimal inspection path; Anomaly assessment is performed on the optimal inspection path to obtain anomaly assessment information, and in response to the anomaly assessment information indicating route feasibility, the optimal inspection path is determined as the UAV inspection path. In response to the abnormal assessment information indicating that the route is not feasible, the path adjustment operation is performed again.

4. The method according to claim 1, wherein, The step of performing target detection on the video stream data to obtain a set of images of illegally parked vehicles includes: Spatiotemporal feature encoding is performed on the video stream data to construct a spatiotemporal feature pyramid set; Target detection is performed on the spatiotemporal feature pyramid set to obtain an initial vehicle detection box image set; The initial set of vehicle detection box images is correlated across frames to obtain a set of vehicle trajectory segments; Motion state analysis is performed on the set of vehicle trajectory segments to obtain a candidate set of stationary vehicle trajectories; Based on the geographic information network system, spatial overlay analysis is performed on the candidate set of stationary vehicle trajectories to obtain the set of vehicle trajectories in the no-parking area; Based on the preset dwell time and the vehicle trajectory set in the no-parking area, determine the trajectory set of illegally parked vehicles; Keyframes are extracted from the trajectory set of illegally parked vehicles to obtain an image set of illegally parked vehicles.

5. The method according to claim 1, wherein, The step of digitally encrypting the illegally parked vehicle image set, the vehicle location dataset, and the system timestamp to obtain an illegal parking evidence file includes: Based on the preset evidence collection standards, the vehicle location dataset, and the system timestamp, evidence description information is generated; The evidence description information is converted into a binary encoded stream; The binary encoded stream is encrypted with a digital watermark to generate a digital watermark signal set; The digital watermark signal set is embedded into the illegally parked vehicle image set to obtain the illegally parked image set; The illegal parking image set and the evidence description information are associated and encapsulated to obtain the illegal parking evidence file.

6. The method according to claim 1, wherein, The step of controlling associated drones to collect video data of no-parking areas according to the drone inspection path, and obtaining video stream data, includes: Based on the UAV inspection path, a flight waypoint sequence and hovering point information set are generated; Based on the flight waypoint sequence, a flight control command set is generated, and the flight control command set is sent to the UAV flight control system to control the UAV to fly and hover; In response to the drone arriving at any hovering point coordinate corresponding to the hovering point information set, the drone is controlled to enter a hovering state. In response to the drone entering a hovering state, the drone is controlled to perform video acquisition according to a preset frame rate and resolution to obtain video stream data.

7. A traffic vehicle capture device based on unmanned aerial vehicles, comprising: The unit is configured to determine the UAV inspection path based on a geographic information network system; The control unit is configured to control an associated drone to capture video of a no-parking area according to the drone inspection path, thereby obtaining video stream data; The detection unit is configured to perform target detection on the video stream data to obtain a set of images of illegally parked vehicles; The conversion unit is configured to acquire the real-time positioning data and camera operation parameters of the UAV, and to perform spatial coordinate transformation on the real-time positioning data and camera operation parameters based on the spatial coordinate transformation method and the image set of illegally parked vehicles to obtain a vehicle location dataset. The encryption unit is configured to digitally encrypt the set of images of illegally parked vehicles, the set of vehicle location data, and the system timestamp to obtain an evidence file of illegal parking. The transmission unit is configured to transmit the illegal parking evidence file to the traffic enforcement system, and to trigger a road network warning in response to the traffic enforcement system receiving the illegal parking evidence file.

8. An electronic device, comprising: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-6.

9. A computer-readable medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-6.

Images