Intelligent parking inspection method and device for improving sensing precision

Abstract

The invention discloses an intelligent parking inspection method and device for improving sensing precision, and relates to the related technical field of parking inspection, and the method comprises the steps: a parking service platform receives K parking space idle instructions; performing parking monitoring coverage analysis to obtain M blind area parking spaces and W covered parking spaces; parking space vacancy verification is carried out, and a static inspection result is output; extracting historical use data to obtain M pieces of parking space use information; performing parking inspection priority analysis, and outputting an unmanned aerial vehicle inspection line; carrying out idle check on the parking spaces in the M blind areas, and outputting a dynamic inspection result; and updating the idle state of the parking space according to the static inspection result and the dynamic inspection result. The technical problems that in the prior art, a parking space state monitoring means is single, a blind area cannot be effectively covered, a detection result is prone to misjudgment, and complex parking environment changes are difficult to adapt are solved, and the technical effects of improving parking space idle verification accuracy and flexibility and improving parking management efficiency and intelligent degree are achieved.

Description

Technical Field

[0001] This application relates to the technical field of parking inspection, and specifically relates to a smart parking inspection method and device for improving perception accuracy. Background Art

[0002] With the acceleration of the urbanization process, the number of automobiles has shown an explosive growth, and the urban parking problem has become increasingly prominent. As an effective means to relieve parking pressure and improve parking management efficiency, the smart parking system is gradually being widely applied. However, accurately perceiving the status of parking spaces has always been a core challenge. Early parking detection technologies, such as geomagnetic detection technology, can sense the presence of vehicles, but have limited perception capabilities for vehicle types and parking space occupancy details; the installation of inductive loop detection methods is complex, the later maintenance cost is high, and it is easily interfered by factors such as ground construction and environmental changes, resulting in a decrease in detection accuracy; traditional camera monitoring solutions have a large number of monitoring blind spots due to fixed viewing angles, and it is difficult to comprehensively and accurately grasp the real-time status of all parking spaces.

[0003] Therefore, in the current related technologies, there are technical problems such as a single means of monitoring the status of parking spaces, ineffective coverage of blind spots, easy misjudgment of detection results, and difficulty in adapting to changes in complex parking environments. Summary of the Invention

[0004] This application provides a smart parking inspection method and device for improving perception accuracy, which solves the technical problems in the prior art such as a single means of monitoring the status of parking spaces, ineffective coverage of blind spots, easy misjudgment of detection results, and difficulty in adapting to changes in complex parking environments, and achieves the technical effects of improving the accuracy, flexibility of parking space occupancy verification, and the efficiency and intelligence level of parking management.

[0005] This application provides a smart parking inspection method for improving perception accuracy. The method includes: the parking service center receives K parking space occupancy instructions; through extracting the K parking space coordinates of the K parking space occupancy instructions for parking monitoring coverage analysis, M blind spot parking spaces and W covered parking spaces are obtained, where K, M, and W are positive integers, and K = M + W; according to the W covered parking spaces, a fixed pan-tilt camera is scheduled for parking space occupancy verification to output a static inspection result; according to the M parking space coordinates of the M blind spot parking spaces, historical usage data is extracted to obtain M parking space usage information; according to the M parking space usage information and the M parking space coordinates, parking inspection priority analysis is performed to output a drone inspection route; according to the drone inspection route, a drone is scheduled to perform occupancy verification on the M blind spot parking spaces to output a dynamic inspection result; according to the static inspection result and the dynamic inspection result, the occupancy status of the parking spaces is updated.

[0006] In a possible implementation, the intelligent parking inspection method for improving perception accuracy further performs the following processing: interacting with parking lot monitoring information, where the parking lot monitoring information includes the layout position features of N fixed pan-tilt cameras and the coverage range features of the N fixed pan-tilt cameras; locally invoking the parking lot design information; performing monitoring coverage simulation based on the parking lot design information and the parking lot monitoring information to obtain the covered parking space information of the N fixed pan-tilt cameras; associatively storing the N camera identifiers of the N fixed pan-tilt cameras and the N covered parking space information to complete the construction of the monitoring coverage association library.

[0007] In a possible implementation, the intelligent parking inspection method for improving perception accuracy further performs the following processing: constructing a parking lot space model based on the parking lot design information; performing monitoring coverage simulation in the parking lot space model according to the layout position features of the N fixed pan-tilt cameras and the coverage range features of the N fixed pan-tilt cameras to obtain the monitoring coverage ranges of the N fixed pan-tilt cameras; traversing the parking lot space model with the monitoring coverage ranges of the N fixed pan-tilt cameras to perform parking space line coverage verification to obtain the covered parking space information of the N fixed pan-tilt cameras.

[0008] In a possible implementation, the intelligent parking inspection method for improving perception accuracy further performs the following processing: extracting the first parking space coordinates from the first parking space idle instruction; traversing the monitoring coverage association library with the first parking space coordinates to perform parking monitoring coverage analysis and output the first analysis result; if the first analysis result is the first camera identifier, associatively storing the first camera identifier and the first parking space coordinates as the first covered parking space; if the first analysis result is an empty set, taking the first parking space coordinates as the first blind area parking space; and so on, performing parking monitoring coverage analysis according to the K parking space coordinates to obtain the M blind area parking spaces and the W covered parking spaces.

[0009] In a possible implementation, the intelligent parking inspection method for improving perception accuracy further performs the following processing: extracting the first camera identifier and the first parking space coordinates from the first covered parking space; extracting the first layout position feature from the layout position features of the N fixed pan-tilt cameras according to the first camera identifier; fitting and generating the first pan-tilt scheduling strategy according to the first layout position feature and the first parking space coordinates; running the first fixed pan-tilt camera according to the first pan-tilt scheduling strategy to perform the idle verification of the first covered parking space and output the first verification result; and so on, scheduling the fixed pan-tilt cameras according to the W covered parking spaces to perform the idle verification of the parking spaces and output the W verification results, and the W verification results constitute the static inspection results.

[0010] In a possible implementation, the intelligent parking inspection method for improving perception accuracy further performs the following processing: calculating M parking space usage frequencies and M parking space reservation frequencies based on the M parking space usage information; after positioning the M blind spot parking spaces in the parking lot space model according to the M parking space coordinates, calculating M parking space real-time loads based on the spatial characteristics of the M blind spot parking spaces; calculating the weighted values of the M parking space usage frequencies, the M parking space reservation frequencies, and the M parking space real-time loads, and outputting the inspection urgency levels of the M parking spaces; updating the inspection urgency levels of the M parking spaces according to the parking space types of the M blind spot parking spaces, and outputting M updated inspection urgency levels; after sorting the inspection urgencies of the M blind spot parking spaces according to the M updated inspection urgency levels, fitting the shortest inspection route, and outputting the UAV inspection route.

[0011] In a possible implementation, the intelligent parking inspection method for improving perception accuracy further performs the following processing: the parking service middle platform sends the UAV inspection route to the parking lot UAV base station; the parking lot UAV base station performs inspection energy consumption analysis on the UAV inspection route, and matches idle UAVs according to the analysis results to obtain target UAVs; starting the target UAV to traverse the M blind spot parking spaces along the UAV inspection route for idle verification, and outputting the dynamic inspection results.

[0012] This application also provides an intelligent parking inspection device for improving perception accuracy, including: a parking space idle instruction receiving unit, configured to receive K parking space idle instructions from a parking service middle platform; a monitoring coverage analysis unit, configured to perform parking monitoring coverage analysis by extracting the K parking space coordinates of the K parking space idle instructions to obtain M blind spot parking spaces and W covered parking spaces, where K, M, and W are positive integers, and K = M + W; a parking space idle verification unit, configured to schedule a fixed pan-tilt camera according to the W covered parking spaces for parking space idle verification, and output static inspection results; a parking space usage information obtaining unit, configured to extract historical usage data according to the M parking space coordinates of the M blind spot parking spaces to obtain M parking space usage information; an inspection route output unit, configured to perform parking inspection priority analysis according to the M parking space usage information and the M parking space coordinates, and output a UAV inspection route; a dynamic inspection result output unit, configured to schedule a UAV according to the UAV inspection route to perform idle verification on the M blind spot parking spaces, and output dynamic inspection results; a parking space idle state update unit, configured to update the parking space idle state according to the static inspection results and the dynamic inspection results.

[0013] A smart parking inspection method and device for improving perception accuracy proposed in this application. The parking service middle platform receives K parking space free instructions; conducts parking monitoring coverage analysis to obtain M blind spot parking spaces and W covered parking spaces; conducts parking space free verification to output static inspection results; extracts historical usage data to obtain M parking space usage information; conducts parking inspection priority analysis to output the drone inspection route; conducts free verification of the M blind spot parking spaces to output dynamic inspection results; and updates the parking space free status according to the static inspection results and the dynamic inspection results. This solves the technical problems in the prior art such as single means of monitoring parking space status, ineffective coverage of blind spots, easy misjudgment of detection results, and difficulty in adapting to changes in complex parking environments, and achieves the technical effects of improving the accuracy, flexibility of parking space free verification, and the efficiency and intelligence level of parking management. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] To more clearly illustrate the technical solutions of the embodiments of the present disclosure, the accompanying drawings of the embodiments of the present disclosure will be briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the devices according to the embodiments of the application. It should be understood that the operations in the front or below do not necessarily need to be executed precisely in sequence. On the contrary, according to needs, they can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several operations can be removed from these processes.

[0015] Figure 1 Schematic flowchart of a smart parking inspection method for improving perception accuracy provided by an embodiment of the present application.

[0016] Figure 2 Schematic structural diagram of a smart parking inspection device for improving perception accuracy provided by an embodiment of the present application.

[0017] Description of reference numerals: Parking space free instruction receiving unit 10, monitoring coverage analysis unit 20, parking space free verification unit 30, parking space usage information obtaining unit 40, inspection route output unit 50, dynamic inspection result output unit 60, parking space free status updating unit 70. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of this application more obvious and understandable, the following specifically illustrates the specific embodiments of this application.

[0019] To make the objectives, technical solutions and advantages of this application clearer, the following will further describe this application in detail with reference to the accompanying drawings. The described embodiments should not be construed as limitations on this application. All other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application.

[0020] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with each other without conflict. The terms "first / second" involved are only used to distinguish similar objects and do not represent a specific order for the objects. The terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application.

[0021] An embodiment of this application provides a smart parking inspection method for improving perception accuracy, as Figure 1 shown, the method includes:

[0022] Step S100, the parking service middle platform receives K parking space idle instructions.

[0023] Preferably, the parking service middle platform has the functions of data reception, parsing and processing, and can receive and identify these idle instructions from different parking spaces. The parking service middle platform receives K parking space idle instructions. Among them, sensors or detection devices distributed in each parking space, such as geomagnetic sensors, ultrasonic sensors, inductive loops, etc., and intelligent devices related to the parking space, such as intelligent gate barriers, intelligent parking locks, etc., when these devices detect that a vehicle has left the parking space and the parking space is in an idle state, will generate corresponding K parking space idle signals (i.e., parking space idle instructions), which may contain relevant information about the parking space, such as parking space number, geographical location coordinates, etc., and then after encoding, packaging and other processing, they are sent to the parking service middle platform in the form of instructions through a wired network (such as Ethernet) or a wireless network (such as ZigBee, LoRa, NB-IoT, etc.). Here, K is a positive integer representing the number of idle parking spaces.

[0024] Step S200, perform parking monitoring coverage analysis by extracting the K parking space coordinates of the K parking space idle instructions to obtain M blind spot parking spaces and W covered parking spaces, where K, M, and W are positive integers and K = M + W.

[0025] Preferably, among the K parking space availability instructions received by the parking service middleware, each instruction corresponds to a parking space, and each parking space has its specific coordinate information (which can be understood as the position identifier on the parking area map). The parking service middleware extracts the coordinate data of each parking space from these K parking space availability instructions, that is, the K parking space coordinates; then, perform parking monitoring coverage analysis on these K parking space coordinates. Specifically, fixed monitoring devices such as fixed pan-tilt cameras are usually installed in the parking area. The viewing angles and monitoring ranges of these monitoring devices are limited. For the coordinates of each parking space, according to parameters such as the installation position, orientation, and viewing angle range of the fixed monitoring device, determine whether the parking space is within the monitoring coverage of the fixed monitoring device. If the coordinates of a certain parking space are within the monitoring viewing angle range of the fixed monitoring device and can be clearly observed by the fixed monitoring device, then this parking space is identified as a "covered parking space". After analyzing and judging each of the K parking space coordinates one by one, the number of all parking spaces that can be covered by the fixed monitoring device is recorded as W; on the contrary, if the coordinates of a certain parking space are not within the monitoring viewing angle range of the fixed monitoring device and cannot be directly observed by the fixed monitoring device, then this parking space is identified as a "blind zone parking space". Similarly, after analysis, the number of all parking spaces in the monitoring blind zone is recorded as M; because the K parking spaces are composed of parking spaces within the monitoring coverage (i.e., W covered parking spaces) and parking spaces in the monitoring blind zone (i.e., M blind zone parking spaces), so K, M, and W satisfy the relationship K = M + W, and K, M, and W are all positive integers.

[0026] Further, step S200 further includes step S201 of interacting parking lot monitoring information, where the parking lot monitoring information includes the N layout position characteristics and N coverage range characteristics of N fixed pan-tilt cameras; step S202 of locally invoking the parking lot design information; step S203 of performing monitoring coverage simulation according to the parking lot design information and the parking lot monitoring information to obtain the N coverage parking space information of the N fixed pan-tilt cameras; step S204 of associatively storing the N camera identifiers of the N fixed pan-tilt cameras and the N coverage parking space information to complete the construction of the monitoring coverage association library.

[0027] Preferably, the parking lot monitoring information is obtained through interaction with the monitoring equipment, including the layout position characteristics of N fixed PTZ cameras (such as the specific coordinate position, installation height, orientation, etc. of the camera in the parking lot) and coverage characteristics (viewing angle of the camera, horizontal and vertical monitoring range, etc.), wherein N is a positive integer representing the number of fixed PTZ cameras; the parking lot design information includes the overall layout of the parking lot, such as the distribution of parking spaces, the direction of passages, the location of buildings and facilities, etc.; based on the acquired parking lot design information and monitoring information, the monitoring coverage of each fixed PTZ camera is simulated, and in the simulation process, the camera layout position, orientation, coverage, and other parameters are calculated. The number is combined with the parking space layout of the parking lot to determine whether each parking space is within the monitoring range of a certain camera. For example, by calculating the geometric relationship between the camera viewing angle and the parking space position, it is determined which parking spaces can be clearly monitored by the camera, and then the N covered parking space information corresponding to each of the N fixed PTZ cameras is obtained, that is, the list of parking spaces that each camera can monitor; finally, the N camera identifiers (such as unique identifiers such as the camera number and name) of the N fixed PTZ cameras are associated and stored with their corresponding N covered parking space information, so that the management personnel can quickly query which parking spaces can be monitored by a certain camera, or which cameras monitor a certain parking space.

[0028] Furthermore, step S203 also includes step A, constructing a parking lot space model according to the parking lot design information; step B, performing a monitoring coverage simulation in the parking lot space model according to the N layout location features and N coverage range features, and obtaining the N monitoring coverage ranges of the N fixed pan-tilt cameras; and step C, using the N monitoring coverage ranges to traverse the parking lot space model to perform parking space line coverage verification, and obtain the N covered parking space information.

[0029] Preferably, according to the parking lot design information, professional modeling software is used to convert the actual layout of the parking lot into a digital space model, which includes various elements in the parking lot, such as the precise location and size of the parking spaces, the direction and width of the passageways, the locations of buildings and other facilities, and other information. For example, the model will accurately mark the coordinates of the four corner points of each parking space, as well as the relative position relationship between the passageways and the parking spaces; based on the constructed parking lot space model, combined with the N layout position features of the N fixed PTZ cameras (such as the camera's installation coordinates, height, orientation, etc.) and N coverage range features (such as the camera's viewing angle, horizontal and vertical monitoring range, etc.), the monitoring coverage range of each camera is simulated to determine the coverage of the parking lot space. In the space model, each camera can cover the area. For example, according to the installation position and direction of the camera and its viewing angle, the spatial range that can be "seen" from the camera position is calculated, so as to obtain the monitoring coverage of each camera; then use N monitoring coverage ranges to traverse the parking lot space model to verify the parking space line coverage. The verification standard is that when a monitoring coverage range completely includes all the parking space lines of a parking space, it is considered that the monitoring can perform idle verification on the parking space, that is, the parking space belongs to the coverage of this camera. By traversing all the parking spaces in the parking lot space model, it is checked whether the parking space lines of each parking space are completely within the monitoring coverage of a certain camera, and finally the information of N covered parking spaces is obtained.

[0030] Furthermore, step S200 also includes step S210, extracting the first parking space coordinates from the first parking space vacant instruction; step S220, using the first parking space coordinates to traverse the monitoring coverage association library, perform parking monitoring coverage analysis, and output a first analysis result; step S230, if the first analysis result is a first camera identifier, then the first camera identifier is associated with the first parking space coordinates and stored as the first covered parking space; step S240, if the first analysis result is an empty set, then the first parking space coordinates are used as the first blind spot parking space; step S250, and so on, perform parking monitoring coverage analysis according to the K parking space coordinates to obtain the M blind spot parking spaces and W covered parking spaces.

[0031] Preferably, when the parking service middleware receives the first parking space free instruction, the corresponding first parking space coordinates are extracted from the instruction, and the monitoring coverage association library is traversed using the extracted first parking space coordinates. Specifically, during the traversal process, it is checked whether the coverage parking space information of each camera contains the first parking space coordinates, and the first analysis result is output. That is, if the coordinate exists in the coverage parking space information of a certain camera, it means that this parking space is within the monitoring coverage of this camera. If the coordinate is not found after traversing the coverage parking space information of all cameras, it means that this parking space is not within the monitoring coverage of any fixed pan-tilt camera. If, after traversing the monitoring coverage association library, the first analysis result is a first camera identifier, that is, a camera that can cover this parking space is found, then this first camera identifier is associated and stored with the first parking space coordinates, and this parking space is marked as the first covered parking space, indicating that it is within the monitoring coverage of a certain fixed pan-tilt camera, and operations such as detecting the free state of the parking space can be performed through this camera. If the first analysis result is an empty set, that is, no camera that can cover this parking space is found, then the first parking space coordinates are used as the first blind area parking space, indicating that this parking space cannot be directly monitored by the existing fixed pan-tilt cameras, and other methods (such as drone patrol) need to be used to obtain its parking space status information. For each of the K received parking space free instructions, parking monitoring coverage analysis is performed, and finally M blind area parking spaces and W covered parking spaces can be obtained, thus completing the classification of the monitoring coverage status of all parking spaces in the parking lot.

[0032] Step S300, schedule the fixed pan-tilt camera according to the W covered parking spaces for parking space free verification, and output the static patrol result.

[0033] Preferably, based on the location distribution of W covered parking spaces, the positions and monitoring ranges of fixed pan-tilt cameras, the parking service middleware formulates a reasonable scheduling strategy. For example, for adjacent covered parking spaces, the same fixed pan-tilt camera is preferably scheduled for inspection to reduce the rotation and switching times of the camera and improve the inspection efficiency. At the same time, considering the camera's viewing angle and focal length, ensure that the status of each covered parking space can be clearly observed; according to the formulated scheduling strategy, the parking service middleware sends control commands to the corresponding fixed pan-tilt cameras, including parameters such as the rotation angle and zoom ratio of the camera, so that the camera can be aligned with the covered parking space to be inspected; after the fixed pan-tilt camera adjusts its posture according to the command, it collects images of the covered parking space. After the camera obtains the real-time image of the parking space, it analyzes the status of the parking space through image recognition technology, and identifies whether there is a vehicle in the parking space area in the image. The judgment basis may include the vehicle's outline, color characteristics, whether the parking space line is blocked, etc. For example, if there is no vehicle outline in the parking space area in the image and the parking space line is completely visible, it is preliminarily judged that the parking space is in an idle state. Finally, the status information of the parking space after image recognition processing is transmitted back to the parking service middleware, and the parking service middleware integrates this information to form a static inspection result, clearly showing the idle or occupied status of each covered parking space.

[0034] Further, step S300 further includes step S310 of extracting the first camera identifier and the first parking space coordinates from the first covered parking space; step S320 of extracting the first layout position feature from the N layout position features according to the first camera identifier; step S330 of fitting and generating the first pan-tilt scheduling strategy according to the first layout position feature and the first parking space coordinates; step S340 of running the first fixed pan-tilt camera according to the first pan-tilt scheduling strategy to perform the idle verification of the first covered parking space and output the first verification result; step S350, and so on, scheduling the fixed pan-tilt cameras according to the W covered parking spaces to perform the parking space idle verification and outputting W verification results, and the W verification results constitute the static inspection result.

[0035] Preferably, for the first covered parking space, the corresponding first camera identifier and the first parking space coordinates are extracted from the stored association information. The first camera identifier is used to uniquely identify the camera capable of monitoring the parking space, and the first parking space coordinates specify the specific location of the parking space in the parking lot. According to the extracted first camera identifier, in the information set containing the layout position characteristics of N fixed pan-tilt cameras, the first layout position characteristic corresponding to the camera is found and extracted. The first layout position characteristic may include detailed information such as the installation coordinates of the camera (such as the specific X and Y coordinate values in the parking lot plane coordinate system), installation height, and orientation angle. Then, using the obtained first layout position characteristic and the first parking space coordinates, a first pan-tilt scheduling strategy is generated. For example, according to the installation position and orientation of the camera, and the coordinates of the parking space, parameters such as the angles (horizontal and vertical directions) that the camera needs to rotate and whether the focal length needs to be adjusted are calculated to guide how the fixed pan-tilt camera adjusts its own posture to achieve effective monitoring of the specified parking space.

[0036] Preferably, according to the generated first pan-tilt scheduling strategy, the first fixed pan-tilt camera is controlled to operate, so that it adjusts its posture according to the strategy and performs image acquisition or other forms of detection on the first covered parking space. That is, by analyzing and processing the collected data (such as parking space images), the idle state of the parking space is judged, and the first verification result is output. For example, if the image shows that there is no vehicle on the parking space, the first verification result is "idle"; if there is a vehicle, it is "occupied". For each of the W covered parking spaces, the above operations are repeated in sequence: extracting relevant information, obtaining the camera layout position characteristics, generating the pan-tilt scheduling strategy, operating the fixed pan-tilt camera to verify the parking space idle state and output the verification result. Finally, the W verification results are integrated to form a static inspection result, which comprehensively reflects the current idle state of all parking spaces within the monitoring coverage range in the parking lot, thus realizing the accurate monitoring and status detection of the covered parking spaces, and ensuring that the parking lot management system can timely and accurately master the usage of the parking spaces.

[0037] Step S400, extracting historical usage data according to the M parking space coordinates of the M blind area parking spaces to obtain M parking space usage information.

[0038] Preferably, taking the coordinates of M blind spot parking spaces as the key indexes, by traversing the parking management database, the record positions corresponding to these M coordinates are found. Just like in a large warehouse, through specific coordinate tags, the shelf positions where specific items are stored can be found. Each parking space coordinate is unique and can accurately locate all relevant data storage points of the corresponding parking space in the database, ensuring that the extracted data accurately corresponds to these M blind spot parking spaces. Specifically, according to the located positions, historical usage data is extracted, including the date and time records of the parking space being occupied, which can reflect the usage situation of the parking space at different times in the past, understand which dates and time periods the parking space is more frequently used, and which time periods are more idle; it also includes the specific times when each vehicle enters and leaves the parking space. By calculating the difference between the entry and exit times, the parking duration of each vehicle in the blind spot parking space can be obtained; in addition, there may be records of vehicle types. Different types of vehicles may have different usage requirements and usage methods for parking spaces. For example, the parking occupancy situations of small cars and large SUVs may be different; finally, the extracted data of each parking space is sorted out to form corresponding M parking space usage information. For example, for each blind spot parking space, its average parking duration, maximum parking duration, and minimum parking duration are calculated, the parking frequencies in different time periods are counted, and the usage proportions of different vehicle types are analyzed, etc.

[0039] Step S500, based on the M parking space usage information and the M parking space coordinates, conduct an analysis of the parking inspection priority and output the drone inspection route.

[0040] Preferably, determine the factors affecting the priority, including usage frequency, parking duration, and geographical location, etc. Among them, count the number of times each blind spot parking space has been occupied in the past period. A parking space with a high usage frequency means that its status may change more frequently, and it needs to be inspected more timely to ensure the accuracy of the data. For example, a parking space occupied 20 times in a month has a higher inspection priority than a parking space occupied only 5 times; analyze data such as the average parking duration and the longest parking duration of vehicles in each blind spot parking space. If a certain parking space has a relatively short average parking duration and vehicles enter and leave frequently, then its status update may be more urgent. Prioritizing the inspection can reduce management errors caused by information lag; combined with the M parking space coordinates, consider the location of the parking space. Parking spaces located near traffic arteries, around the parking lot entrance and exit and other key areas have a greater impact on the smoothness of the overall parking management, and their priorities are higher; assign weights to each influencing factor, that is, quantify the priority. For example, set the weight of usage frequency as 0.4, the weight of parking duration as 0.3, and the weight of geographical location and surrounding environment as 0.3 (the weight assignment can be adjusted according to actual parking management needs and experience), and score each blind spot parking space according to its performance in each factor.

[0041] Preferably, based on the coordinates of the M parking spaces, a path planning algorithm is used to prioritize the parking spaces with high priority in the inspection route planning. For example, a Dijkstra algorithm is used to find a route that can cover the high-priority blind spot parking spaces in sequence with the shortest flight distance or the shortest time, while meeting the flight safety and performance limitations of the drone (such as cruising range, flight altitude limitations, etc.), to ensure that in the planned inspection route, the drone will not make unnecessary repeated flights to the same area, and avoid conflicts between the inspection routes of different drones (if the parking lot uses multiple drones for inspection). By reasonably setting the flight direction, altitude and sequence, the inspection route is optimized so that the drone can efficiently complete the inspection tasks of the M blind spot parking spaces, and finally output the drone inspection route, which contains detailed flight instruction information, such as the coordinates of each waypoint, flight speed, flight altitude, dwell time (for taking pictures or detection), etc., so as to achieve efficient and targeted dynamic inspection of blind spot parking spaces.

[0042] Furthermore, step S500 also includes step S510, calculating M parking space usage frequencies and M parking space reservation frequencies based on the M parking space usage information; step S520, after locating the M blind spot parking spaces in the parking lot space model according to the M parking space coordinates, calculating the M parking space real-time loads according to the spatial characteristics of the M blind spot parking spaces; step S530, weighted calculation of the M parking space usage frequencies, M parking space reservation frequencies and M parking space real-time loads, and outputting M parking space inspection urgency; step S540, updating the M parking space inspection urgency according to the parking space type of the M blind spot parking spaces, and outputting M updated inspection urgency; step S550, after sorting the M blind spot parking spaces according to the M updated inspection urgency, performing shortest inspection route fitting, and outputting the drone inspection route.

[0043] Preferably, M parking space usage frequencies and M parking space reservation frequencies are calculated based on the M parking space usage information. Specifically, by counting the number of times each parking space is occupied within a specific time period (such as a week, a month, etc., which can be set according to actual management requirements) and then dividing it by the duration of this time period, the M parking space usage frequencies can be obtained. The higher the usage frequency, the more frequent the vehicle flow in this parking space, and the greater the possibility of its state change. When the parking lot has a parking space reservation function and there are corresponding reservation data, by sorting out these data and counting the number of times each blind spot parking space is reserved within a specific future time period (such as the next week), the M parking space reservation frequencies can be obtained, reflecting the future demand heat of the parking spaces. According to the M parking space coordinates, the M blind spot parking spaces are located in the parking lot space model, that is, the specific positions of each blind spot parking space are accurately located in the model. Each blind spot parking space has its unique spatial characteristics, including the area size of the parking space, the distance from surrounding parking spaces, the width of the channels in the area where it is located, etc. Considering these factors comprehensively, the real-time load of the parking space is calculated. For example, if a parking space has a large area, wide spacing from surrounding parking spaces, and a relatively spacious channel, then its ability to accommodate vehicle entry and exit per unit time is relatively strong, and the real-time load is relatively low; on the contrary, if the parking space has a small area, dense surrounding parking spaces, and a narrow channel, making it difficult for vehicles to enter and exit, its real-time load is relatively high. Thus, the real-time load values of the M parking spaces are obtained, reflecting the current tightness of each blind spot parking space in terms of space utilization. Parking spaces with a high real-time load are more likely to have parking-related problems and require more frequent inspections.

[0044] Preferably, according to the actual management focus and experience of the parking lot, different weights are assigned to the parking space usage frequency, reservation frequency, and real-time load for the inspection urgency respectively. For example, assuming the weight of the parking space usage frequency is 0.4, the weight of the parking space reservation frequency is 0.3, and the weight of the parking space real-time load is 0.3. For each blind spot parking space, a weighted calculation is performed, that is, the corresponding parking space usage frequency, reservation frequency, and real-time load values are multiplied by their respective weights and then added together to obtain the inspection urgency of the parking space. For different types of parking spaces, such as ordinary parking spaces, disabled-person special parking spaces, VIP parking spaces, etc., their importance and management requirements are different. According to the parking space types of the M blind spot parking spaces, the inspection urgencies of the M parking spaces calculated previously are adjusted to ensure that these special parking spaces are always available and in accurate status, and then M updated inspection urgencies are output, making the division of inspection priorities more in line with the actual management needs of the parking lot. Then, according to the M updated inspection urgencies, the M blind spot parking spaces are sorted, and the parking spaces with high inspection urgency are ranked in the front, ensuring that in the case of limited resources (such as limited UAV battery power and flight time), the parking spaces that most need to be inspected are inspected first. Finally, using a path planning algorithm (such as the Dijkstra algorithm, etc.), combined with the actual layout of the parking lot (reflected in the parking lot space model) and the flight performance limitations of the UAV (such as maximum flight distance, flight speed, turning radius, etc.), the shortest inspection route is fitted, and then the UAV inspection route is output, which can not only ensure the inspection of all blind spot parking spaces but also save the flight time and battery power of the UAV to the greatest extent and improve the inspection efficiency.

[0045] Step S600, dispatch the UAV according to the UAV inspection route to perform the idle check of the M blind spot parking spaces, and output the dynamic inspection result.

[0046] Preferably, the parking service middleware sends instructions to the corresponding drone according to the determined drone inspection route, including information such as the take-off location, the coordinates of each waypoint on the flight path, the flight speed, and the flight altitude. The drone is dispatched to fly to the corresponding positions above the M blind spot parking spaces in sequence according to the planned inspection route. Then, the detection equipment carried by itself (such as a high-definition camera, an infrared sensor, etc.) is used to check and detect the parking spaces. Taking the camera as an example, the drone will stay at an appropriate height and angle above the parking space for a period of time, take a high-definition image of the parking space, and analyze the captured image through image recognition technology to determine whether there is a vehicle in the parking space. Specifically, the object features in the parking space area of the image are identified and compared with the feature model of the vehicle. If an object that conforms to the vehicle features is found, it is determined that the parking space is occupied; if there is no object with vehicle features in the parking space area, it is determined that the parking space is in an idle state. In addition, the drone can be equipped with an infrared sensor to assist in judging whether the parking space is occupied by detecting information such as temperature changes in the parking space area. After the drone completes the detection of all M blind spot parking spaces, it will transmit the detection results of each parking space back to the parking service middleware, and then generate the verification information of the M blind spot parking spaces, which is output as the dynamic inspection result to ensure the accuracy of the parking space idle state information.

[0047] Further, step S600 further includes step S610, where the parking service middleware sends the drone inspection route to the parking lot drone base station; step S620, the parking lot drone base station conducts an inspection energy consumption analysis on the drone inspection route, and matches idle drones according to the analysis result to obtain the target drone; step S630, starts the target drone to traverse the M blind spot parking spaces along the drone inspection route for idle verification, and outputs the dynamic inspection result.

[0048] Preferably, after the parking service middleware completes the analysis of the M blind spot parking spaces and generates the drone inspection route, it sends this inspection route to the parking lot drone base station. Among them, the inspection route includes the coordinates of each waypoint that the drone needs to fly to in sequence (corresponding to the positions of the M blind spot parking spaces and possible intermediate transition positions), as well as relevant flight parameters, such as flight speed, flight altitude, etc. information; then the parking lot drone base station receives the drone inspection route and conducts an inspection energy consumption analysis on this route. Specifically, considering multiple factors such as the distance of the drone flying from one waypoint to the next, the change in flight altitude, and the setting of flight speed, according to the power system parameters of the drone (such as battery capacity, power consumption per unit distance, etc.) and these flight parameters, calculate the approximate energy consumption required for the drone to complete the inspection task along this inspection route. For example, if the inspection route is relatively long and includes multiple waypoints that require frequent changes in flight direction and altitude, then the expected energy consumption will be relatively high.

[0049] Preferably, the base station matches the idle drones in the parking lot according to the analysis results, that is, queries the list of currently idle drones, obtains information such as the battery power and flight performance of each idle drone, and then compares the estimated inspection energy consumption with the battery power of each idle drone, and selects the drones with sufficient battery power and meeting the flight performance requirements of the inspection task (such as maximum flight speed, flight altitude range, etc.) as the target drones; finally, the parking lot drone base station starts the drone and sends the drone inspection route to the target drone. According to the received inspection route, it flies over M blind spot parking spaces in turn, and uses the detection equipment carried by itself (such as high-definition cameras, infrared sensors, etc.) to perform idle verification on the parking spaces. For example, by taking high-definition images of the parking spaces and performing image recognition, it is judged whether there is a vehicle on the parking space. After completing the detection of all M blind spot parking spaces, the drone transmits the detection results of each parking space to the parking lot drone base station, and the base station further sends these results to the parking service middle platform. The parking service middle platform sorts and outputs the dynamic inspection results, realizing the effective inspection of the blind spot parking spaces in the parking lot and ensuring the real-time accuracy of the parking space status information in the parking lot.

[0050] Step S700, update the idle state of the parking space according to the static inspection result and the dynamic inspection result.

[0051] Preferably, compare the status of the parking spaces covered in the static inspection with the status of the parking spaces in the same area that may be involved in the dynamic inspection process. If the two inspection results are consistent, for example, both the static and dynamic inspections show that a certain parking space is in an idle state, then it can be preliminarily determined that the idle state of this parking space is accurate; if there is an inconsistent situation, for example, the static inspection shows that the parking space is idle, while the dynamic inspection finds that there is a vehicle parked, then it is necessary to further analyze the reason, which may be that there is a delay in the static inspection or an error in the dynamic inspection; for blind spot parking spaces, since the static inspection cannot directly obtain their status, the idle state is mainly determined based on the dynamic inspection results, but it is also necessary to make a comprehensive judgment in combination with the overall usage rules and historical data of the parking lot. For example, if the dynamic inspection shows that a certain blind spot parking space is idle, but according to historical data, this parking space is usually occupied during this period, then it is necessary to verify again, or mark the status of this parking space as to be confirmed, waiting for the next inspection or other information to further determine; according to the results of the comparison and analysis, update the parking space status database of the parking lot, accurately record the latest idle state of each parking space in the database, and ensure that the information in the database is consistent with the actual parking space status. For example, if it is determined through comprehensive judgment that a certain parking space changes from idle to occupied, the status field of this parking space in the database is updated from "idle" to "occupied".

[0052] In the above text, refer to Figure 1A method for intelligent parking inspection to improve perception accuracy according to an embodiment of the present invention is described in detail. Next, reference will be made to Figure 2 Describe an intelligent parking inspection device for improving perception accuracy according to an embodiment of the present invention.

[0053] An intelligent parking inspection device for improving perception accuracy according to an embodiment of the present invention is used to solve the technical problems existing in the prior art, such as single means for monitoring the parking space status, ineffective coverage of blind spots, easy misjudgment of detection results, and difficulty in adapting to changes in complex parking environments. It achieves the technical effects of improving the accuracy, flexibility of parking space idle verification, and the efficiency and intelligence level of parking management. As Figure 2 shown, an intelligent parking inspection device for improving perception accuracy includes: a parking space idle instruction receiving unit 10, a monitoring coverage analysis unit 20, a parking space idle verification unit 30, a parking space usage information obtaining unit 40, a patrol route output unit 50, a dynamic patrol result output unit 60, and a parking space idle status updating unit 70.

[0054] The parking space idle instruction receiving unit 10 is used for the parking service middleware to receive K parking space idle instructions; the monitoring coverage analysis unit 20 is used for performing parking monitoring coverage analysis by extracting the K parking space coordinates of the K parking space idle instructions to obtain M blind spot parking spaces and W covered parking spaces, where K, M, and W are positive integers, and K = M + W; the parking space idle verification unit 30 is used for scheduling a fixed pan-tilt camera according to the W covered parking spaces for parking space idle verification and outputting a static patrol result; the parking space usage information obtaining unit 40 is used for extracting historical usage data according to the M parking space coordinates of the M blind spot parking spaces to obtain M parking space usage information; the patrol route output unit 50 is used for performing parking patrol priority analysis according to the M parking space usage information and the M parking space coordinates and outputting a drone patrol route; the dynamic patrol result output unit 60 is used for scheduling a drone according to the drone patrol route to perform idle verification of the M blind spot parking spaces and outputting a dynamic patrol result; the parking space idle status updating unit 70 is used for updating the parking space idle status according to the static patrol result and the dynamic patrol result.

[0055] Next, the specific configuration of the monitoring coverage analysis unit 20 will be described in detail. The monitoring coverage analysis unit 20 further includes: interacting with parking lot monitoring information, where the parking lot monitoring information includes the N layout position characteristics and N coverage range characteristics of N fixed pan-tilt cameras; locally invoking parking lot design information; performing monitoring coverage simulation according to the parking lot design information and the parking lot monitoring information to obtain the N covered parking space information of the N fixed pan-tilt cameras; associatively storing the N camera identifiers of the N fixed pan-tilt cameras and the N covered parking space information to complete the construction of the monitoring coverage association library.

[0056] Next, the specific configuration of the monitoring coverage analysis unit 20 will be further described in detail. The monitoring coverage analysis unit 20 further includes: constructing a parking lot space model according to the parking lot design information; performing monitoring coverage simulation in the parking lot space model according to the N layout position features and N coverage range features to obtain the N monitoring coverage ranges of the N fixed pan-tilt cameras; traversing the parking lot space model with the N monitoring coverage ranges to perform a parking space line coverage check, and obtaining the N covered parking space information.

[0057] Next, the specific configuration of the monitoring coverage analysis unit 20 will be further described in detail. The monitoring coverage analysis unit 20 further includes: extracting a first parking space coordinate from the first parking space idle instruction; traversing the monitoring coverage association library with the first parking space coordinate to perform parking monitoring coverage analysis, and outputting a first analysis result; if the first analysis result is a first camera identifier, associating and storing the first camera identifier and the first parking space coordinate as a first covered parking space; if the first analysis result is an empty set, taking the first parking space coordinate as a first blind area parking space; and so on, performing parking monitoring coverage analysis according to the K parking space coordinates to obtain the M blind area parking spaces and W covered parking spaces.

[0058] Next, the specific configuration of the parking space idle check unit 30 will be described in detail. The parking space idle check unit 30 further includes: extracting the first camera identifier and the first parking space coordinate from the first covered parking space; extracting a first layout position feature from the N layout position features according to the first camera identifier; fitting and generating a first pan-tilt scheduling strategy according to the first layout position feature and the first parking space coordinate; operating the first fixed pan-tilt camera according to the first pan-tilt scheduling strategy to perform an idle check of the first covered parking space, and outputting a first check result; and so on, scheduling the fixed pan-tilt cameras according to the W covered parking spaces to perform parking space idle checks, and outputting W check results, and the W check results constitute the static inspection result.

[0059] Next, the specific configuration of the inspection route output unit 50 will be described in detail. The inspection route output unit 50 further includes: calculating M parking space usage frequencies and M parking space reservation frequencies according to the M parking space usage information; after positioning the M blind spot parking spaces in the parking lot space model according to the M parking space coordinates, calculating M parking space real-time loads according to the spatial characteristics of the M blind spot parking spaces; calculating the M parking space usage frequencies, M parking space reservation frequencies, and M parking space real-time loads by weighted calculation, and outputting M parking space inspection urgencies; updating the M parking space inspection urgencies according to the parking space types of the M blind spot parking spaces, and outputting M updated inspection urgencies; after sorting the M blind spot parking spaces according to the M updated inspection urgencies, performing the shortest inspection route fitting, and outputting the UAV inspection route.

[0060] Next, the specific configuration of the dynamic inspection result output unit 60 will be described in detail. The dynamic inspection result output unit 60 further includes: the parking service middleware sending the UAV inspection route to the parking lot UAV base station; the parking lot UAV base station performing inspection energy consumption analysis on the UAV inspection route, and performing idle UAV matching according to the analysis result to obtain a target UAV; starting the target UAV to traverse the M blind spot parking spaces along the UAV inspection route for idle verification, and outputting the dynamic inspection result.

[0061] The intelligent parking inspection device for improving perception accuracy provided by the embodiments of the present invention can execute the intelligent parking inspection method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0062] Although the present application makes various references to certain modules in the device according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The included individual units and modules are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of the functional units are only for the convenience of mutual distinction and do not limit the protection scope of the present invention.

[0063] The above specific embodiments do not constitute a limitation to the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of the present application should be included within the protection scope of the present application.

Claims

1. A smart parking inspection method for improving perception accuracy, characterized in that, The method includes: The parking service middleware receives K parking space free instructions; By extracting the K parking space coordinates of the K parking space free instructions for parking monitoring coverage analysis, M blind area parking spaces and W covered parking spaces are obtained, where K, M, and W are positive integers, and K = M + W; According to the W covered parking spaces, a fixed pan-tilt camera is scheduled for parking space free verification, and a static inspection result is output; According to the M parking space coordinates of the M blind area parking spaces, historical usage data is extracted to obtain M parking space usage information; According to the M parking space usage information and the M parking space coordinates, a parking inspection priority analysis is performed, and a drone inspection route is output; According to the drone inspection route, a drone is scheduled to perform free verification of the M blind area parking spaces, and a dynamic inspection result is output; According to the static inspection result and the dynamic inspection result, the free status of the parking space is updated.

2. The intelligent parking inspection method for improving perception accuracy according to claim 1, wherein Before obtaining M blind area parking spaces and W covered parking spaces by extracting the K parking space coordinates of the K parking space free instructions for parking monitoring coverage analysis, the method further includes: Interact with the parking lot monitoring information, where the parking lot monitoring information includes the N layout position characteristics and the N coverage range characteristics of N fixed pan-tilt cameras; Locally call the parking lot design information; According to the parking lot design information and the parking lot monitoring information, a monitoring coverage simulation is performed to obtain the N covered parking space information of the N fixed pan-tilt cameras; Associate and store the N camera identifiers of the N fixed pan-tilt cameras with the N covered parking space information to complete the construction of the monitoring coverage association library.

3. The intelligent parking inspection method for improving sensing accuracy according to claim 2, characterized in that, According to the parking lot design information and the parking lot monitoring information, a monitoring coverage simulation is performed to obtain the N covered parking space information of the N fixed pan-tilt cameras, and the method includes: Construct a parking lot space model according to the parking lot design information; According to the N layout position characteristics and the N coverage range characteristics, a monitoring coverage simulation is performed in the parking lot space model to obtain the N monitoring coverage ranges of the N fixed pan-tilt cameras; Use the N monitoring coverage ranges to traverse the parking lot space model for parking space line coverage verification to obtain the N covered parking space information.

4. The intelligent parking inspection method for improving sensing accuracy according to claim 2, wherein, When obtaining M blind area parking spaces and W covered parking spaces by extracting the K parking space coordinates of the K parking space free instructions for parking monitoring coverage analysis, the method includes: Extract the first parking space coordinate from the first parking space free instruction; Use the first parking space coordinate to traverse the monitoring coverage association library for parking monitoring coverage analysis, and output a first analysis result; If the first analysis result is the first camera identifier, then associate and store the first camera identifier with the first parking space coordinate as the first covered parking space; If the first analysis result is an empty set, then use the first parking space coordinate as the first blind area parking space; And so on, according to the K parking space coordinates, parking monitoring coverage analysis is performed to obtain the M blind area parking spaces and W covered parking spaces.

5. The intelligent parking inspection method for improving perception accuracy according to claim 4, wherein When scheduling a fixed pan-tilt camera according to the W covered parking spaces for parking space free verification and outputting a static inspection result, the method includes: Extract the first camera identifier and the first parking space coordinates from the first covered parking space; Extract the first layout position feature from the N layout position features according to the first camera identifier; Generate a first pan-tilt scheduling strategy by fitting the first layout position feature and the first parking space coordinates; Run the first fixed pan-tilt camera according to the first pan-tilt scheduling strategy to perform the idle check of the first covered parking space, and output the first check result; And so on, schedule the fixed pan-tilt cameras according to the W covered parking spaces to perform the parking space idle check, and output W check results, and the W check results constitute the static inspection result.

6. The intelligent parking inspection method for improving sensing accuracy according to claim 3, wherein Perform parking inspection priority analysis according to the M parking space usage information and the M parking space coordinates, and output the UAV inspection route. The method includes: Calculate M parking space usage frequencies and M parking space reservation frequencies according to the M parking space usage information; After positioning the M blind area parking spaces in the parking lot space model according to the M parking space coordinates, calculate the real-time load of the M parking spaces according to the spatial characteristics of the M blind area parking spaces; Calculate the M parking space usage frequencies, the M parking space reservation frequencies and the real-time load of the M parking spaces by weighted calculation, and output the inspection urgency of the M parking spaces; Update the inspection urgency of the M parking spaces according to the parking space types of the M blind area parking spaces, and output the updated inspection urgency of the M parking spaces; After sorting the inspections of the M blind area parking spaces according to the updated inspection urgency of the M parking spaces, perform the shortest inspection route fitting, and output the UAV inspection route.

7. The intelligent parking inspection method for improving perception accuracy according to claim 1, wherein Schedule the UAV according to the UAV inspection route to perform the idle check of the M blind area parking spaces, and output the dynamic inspection result. The method includes: The parking service middle platform sends the UAV inspection route to the parking lot UAV base station; The parking lot UAV base station performs inspection energy consumption analysis on the UAV inspection route, and matches the idle UAVs according to the analysis results to obtain the target UAV; Start the target UAV to traverse the idle check of the M blind area parking spaces along the UAV inspection route, and output the dynamic inspection result.

8. An intelligent parking inspection device for improving perception accuracy, characterized in that, The device is used to implement the intelligent parking inspection method for improving the perception accuracy according to any one of claims 1 to 7. The device includes: A parking space idle instruction receiving unit, which is used for the parking service middle platform to receive K parking space idle instructions; A monitoring coverage analysis unit, which is used to perform parking monitoring coverage analysis by extracting the K parking space coordinates of the K parking space idle instructions, and obtain M blind area parking spaces and W covered parking spaces, where K, M, and W are positive integers, and K = M + W; A parking space idle check unit, which is used to schedule the fixed pan-tilt cameras according to the W covered parking spaces to perform the parking space idle check, and output the static inspection result; A parking space usage information obtaining unit, which is used to extract historical usage data according to the M parking space coordinates of the M blind area parking spaces to obtain M parking space usage information; An inspection route output unit, which is used to perform parking inspection priority analysis according to the M parking space usage information and the M parking space coordinates, and output the UAV inspection route; The dynamic inspection result output unit is used to schedule the UAV according to the UAV inspection route to perform the idle check of the M blind spot parking spaces and output the dynamic inspection result; The parking space idle state update unit is used to update the parking space idle state according to the static inspection result and the dynamic inspection result.

Images