Key point posture tracking and early warning method and system in parking process

Abstract

The application relates to the technical field of intelligent parking, and provides a key point posture tracking and early warning method and system in a parking process. The method comprises the following steps: acquiring spatial distribution data information of a target parking lot; performing spatial modeling on the spatial distribution data information to generate a parking space scene model; acquiring a key point set of a target vehicle; acquiring a vehicle key point posture data stream; performing data filtering on the vehicle key point posture data stream; performing feature tracking analysis on a standard vehicle key point posture data stream to obtain key point posture tracking feature information; and when the key point posture tracking feature information exceeds a posture change threshold value, an abnormal early warning module is used to perform abnormal early warning on a parking process posture. The application solves the technical problem that in the prior art, a vehicle is prone to colliding with other vehicles when parking, causing economic losses and affecting the parking order, and achieves the technical effects of improving the safety of vehicle parking and maintaining the parking order.

Description

Technical Field

[0001] This application relates to the field of smart parking technology, and in particular to a method and system for tracking and warning the attitude of key points during parking. Background Technology

[0002] Keypoints are a more abstract concept. In image processing, they generally refer to points that are important for analyzing the problem. When extracting keypoints, edges should be an important reference, but certainly not the only one. For a given object, keypoints should be points that express certain features, not just edge points. Any point that is helpful in analyzing a specific problem can be called a keypoint.

[0003] Nowadays, almost every household owns a vehicle. However, vehicles are prone to collisions and scratches with other vehicles during parking, resulting in economic losses. By extracting the attitude of key points during parking and setting up tracking and early warning methods, the safety of vehicles when parking can be improved.

[0004] In summary, existing technologies have the problem that parked vehicles are prone to colliding with other vehicles, causing economic losses and disrupting parking order. Summary of the Invention

[0005] Therefore, it is necessary to provide a method and system for key point attitude tracking and early warning during parking, which can improve the safety of vehicle parking and maintain parking order, in order to address the above-mentioned technical problems.

[0006] In a first aspect, this application provides a method for key point attitude tracking and early warning during parking. The method includes: acquiring spatial distribution data information of a target parking lot; using 3D modeling technology to spatially model the spatial distribution data information to generate a parking space scene model; extracting key points of a target vehicle based on the parking space scene model to obtain a set of key points of the target vehicle; installing an attitude sensor on the target vehicle and monitoring the set of key points of the target vehicle in real time through the attitude sensor to obtain a vehicle key point attitude data stream; filtering the vehicle key point attitude data stream to generate a standard vehicle key point attitude data stream; performing feature tracking analysis on the standard vehicle key point attitude data stream to obtain key point attitude tracking feature information; setting an attitude change threshold based on parking experience, and when the key point attitude tracking feature information exceeds the attitude change threshold, using an anomaly early warning module to provide an anomaly early warning for the parking process attitude.

[0007] Secondly, this application provides a key point attitude tracking and early warning system during parking, the system comprising: a spatial distribution data information acquisition module, which acquires spatial distribution data information of a target parking lot; a parking space scene model generation module, which uses 3D modeling technology to perform spatial modeling on the spatial distribution data information to generate a parking space scene model; a target vehicle key point set acquisition module, which extracts key points of the target vehicle based on the parking space scene model to acquire a target vehicle key point set; and a vehicle key point data stream acquisition module, which installs an attitude sensor on the target vehicle and, through... The attitude sensor monitors the set of key points of the target vehicle in real time to acquire the attitude data stream of the vehicle key points; the standard vehicle key point attitude data stream generation module is used to filter the vehicle key point attitude data stream to generate a standard vehicle key point attitude data stream; the key point attitude tracking feature information acquisition module is used to perform feature tracking analysis on the standard vehicle key point attitude data stream to obtain key point attitude tracking feature information; the anomaly warning module is used to set an attitude change threshold based on parking experience, and when the key point attitude tracking feature information exceeds the attitude change threshold, the anomaly warning module provides an anomaly warning for the attitude during the parking process.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] First, spatial distribution data of the target parking lot is acquired. Second, 3D modeling technology is used to spatially model the spatial distribution data, generating a parking space scene model. Then, based on the parking space scene model, key points of the target vehicle are extracted to obtain a set of key points for the target vehicle. Next, an attitude sensor is installed on the target vehicle, and the set of key points of the target vehicle is monitored in real time to obtain a vehicle key point attitude data stream. Then, the vehicle key point attitude data stream is filtered to generate a standard vehicle key point attitude data stream. Next, feature tracking analysis is performed on the standard vehicle key point attitude data stream to obtain key point attitude tracking feature information. Finally, an attitude change threshold is set based on parking experience. When the key point attitude tracking feature information exceeds the attitude change threshold, an anomaly warning module is used to provide an anomaly warning for the parking process attitude. This application solves the technical problem in the prior art where vehicles easily collide with other vehicles when parked, causing economic losses and affecting parking order, and achieves the technical effect of improving vehicle parking safety and maintaining parking order.

[0010] The above description is merely an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating a key point attitude tracking and early warning method during berthing in one embodiment;

[0012] Figure 2 This is a flowchart illustrating the process of determining key point attitude tracking feature information in a key point attitude tracking and early warning method during berthing in one embodiment.

[0013] Figure 3 This is a structural block diagram of a key point attitude tracking and early warning system during berthing in one embodiment.

[0014] Figure labeling: 11 Spatial distribution data information acquisition module, 12 Parking space scene model generation module, 13 Target vehicle key point set acquisition module, 14 Vehicle key point data stream acquisition module, 15 Standard vehicle key point attitude data stream generation module, 16 Key point attitude tracking feature information acquisition module, 17 Anomaly warning module. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0016] like Figure 1 As shown, this application provides a method for key point attitude tracking and early warning during berthing, characterized in that the method includes:

[0017] Obtain spatial distribution data of the target parking lot;

[0018] Keypoints, also known as points of interest, are stable and distinctive sets of points on 2D images, 3D point clouds, or curved surface models that can be obtained by defining detection criteria. Technically, the number of keypoints is much smaller than that of the original point cloud or image. They are combined with local feature descriptors to form keypoint descriptors, which are often used to describe a compact representation of the original data, thus accelerating subsequent recognition. This application achieves the technical effect of maintaining traffic order by extracting and tracking keypoints of vehicles during parking, thereby determining whether the vehicle has stopped normally.

[0019] The target parking lot refers to a parking lot arbitrarily selected by staff from multiple parking locations for study; spatial distribution data refers to the spatial distribution of various facilities within the target parking lot, such as the location information of parking spaces and the location distribution information of parking lanes. By identifying the target parking lot and acquiring its spatial distribution data, data support is provided for the subsequent generation of a parking space spatial scene model.

[0020] Obtain parking lot construction and design information through the parking lot management system;

[0021] CCD image sensors are deployed in the target parking lot to collect images of the parking space occupancy status and obtain parking space image information.

[0022] Feature analysis is performed on the parking space image information to obtain the characteristic information of vehicles occupying the parking spaces;

[0023] Based on the parking lot construction and design information, the parking lot space image information, and the vehicle characteristic information of the parking space occupancy, the spatial distribution data information is determined.

[0024] A parking management system is a network system built using computers, network equipment, and lane management equipment to manage vehicle entry and exit, traffic flow guidance, and parking fee collection within a parking lot. It is an essential tool for professional parking lot management companies. By collecting and recording vehicle entry and exit records and their locations within the lot, it achieves comprehensive dynamic and static management of vehicle entry and exit and vehicles within the lot. The parking management system also obtains construction and design information about the parking lot, including its structural dimensions, parking space distribution, and dimensions. A CCD image sensor is a monitoring software that converts light into electrical charge, which is then converted into a digital signal by a chip. This digital signal is compressed and stored in the camera's internal flash memory. The data is transmitted to the computer, and the image is modified as needed. The CCD image sensor acquires images of the parking space occupancy status of the target parking lot, obtaining parking space image information to determine if there are any vehicles parked in the spaces. Feature analysis of the parking space image information involves color and structural analysis to identify feature points. These feature points are then integrated to obtain vehicle occupancy feature information, i.e., the characteristics of vehicles parked in the spaces. Based on the parking lot construction and design information, the parking space image information, and the vehicle occupancy feature information, the spatial distribution data information is determined. Determining this spatial distribution data information lays the groundwork for subsequently constructing a parking space scene model.

[0025] A color space is constructed, and color matching and mapping are performed based on the color space and the parking space image information to determine the vehicle body color feature information;

[0026] The structure extraction calculation is performed on the parking space image information to obtain structural size feature information;

[0027] The structural dimension feature information and the vehicle body color feature are fused to obtain the parking space occupancy vehicle feature information.

[0028] Color space refers to the most commonly used method of representing color information. Its purpose is to describe color in a generally acceptable way under certain standards, such as RGB and YIQ. Color matching and mapping are performed based on the color space and the parking space image information. That is, the color feature information of the vehicle body in the parking space image information is determined according to the color space description of the parking space image information. Structural extraction calculation of the parking space image information refers to extracting the structure of the vehicle in the parking space image information, that is, finding the structure of the parked vehicle from the parking space image information and calculating it to obtain the structural dimension feature information of the vehicle. The structural dimension feature information and the vehicle body color feature are fused to obtain the feature information of the vehicle occupying the parking space. For example, the feature of the vehicle occupying the parking space is a red vehicle with a length of 3800mm, a width of 1600mm, and a height of 1500mm. Obtaining the feature information of the vehicle occupying the parking space contributes to subsequent spatial modeling.

[0029] The spatial distribution data information is spatially modeled using 3D modeling technology to generate a berth space scene model.

[0030] 3D modeling technology refers to the use of 3D production software to construct models with 3D data in a virtual 3D space. In this application, it refers to using 3D production software such as 3DS Max and Maya to perform spatial modeling in a virtual 3D space, combined with the spatial distribution data information, to generate a parking space scene model. This parking space scene model is a model built based on the data information of the target parking lot, including the parking lot construction design, parking space images, and vehicle characteristics of occupied spaces. Generating a parking space scene model by performing spatial modeling on the spatial distribution data information provides support for subsequent key point extraction.

[0031] Acquire modeling data feature information, which includes feature type, spatial distribution location, and structural dimensions;

[0032] Based on the modeling data feature information, the spatial distribution data information is classified and labeled to obtain spatial distribution data feature information;

[0033] The spatial distribution data information is modeled based on the spatial distribution data feature information using 3D modeling technology to generate the berth space scene model.

[0034] The modeling data feature information includes feature type, spatial distribution location, and structural dimensions. Feature type refers to berths, vehicles, etc., spatial distribution location refers to the location information of the aforementioned features, and structural dimensions refer to the size and dimensions of the features. Based on the modeling data feature information, the spatial distribution data information is classified and labeled. Classification and labeling are important components of modern document classification methods, indicating the location of categories, the order of categories, and their interrelationships. In this application, it refers to classifying the spatial distribution data information and labeling and identifying the classified spatial distribution data information to obtain spatial distribution data feature information. Using 3D modeling technology, the spatial distribution data information is modeled based on the spatial distribution data feature information to generate the berth space scene model. The generation of the berth space scene model contributes to the subsequent extraction of vehicle keypoint poses.

[0035] Based on the parking space scene model, key points of the target vehicle are extracted to obtain a set of key points of the target vehicle.

[0036] Key points refer to critical factors that may affect the berthing result or efficiency during the berthing process, such as the vehicle's body angle. By identifying the key factors affecting the berthing result and efficiency, the most influential key points are extracted and integrated to obtain a target vehicle key point set. Obtaining this target vehicle key point set lays the groundwork for subsequently generating a standard vehicle key point attitude data stream.

[0037] An attitude sensor is installed on the target vehicle, and the attitude sensor is used to monitor the set of key points of the target vehicle in real time to obtain the attitude data stream of the vehicle key points.

[0038] An attitude sensor is a sensor used to measure the attitude of an object. It can detect the object's acceleration and angular velocity, and calculate the object's direction and angle. The attitude sensor is used to monitor the set of key points of the target vehicle in real time, acquiring a vehicle key point attitude data stream. The data stream is an ordered sequence of bytes with a start and end point, including an input stream and an output stream. In this application, the vehicle key point attitude data stream refers to the change of the vehicle key points from their initial position to their final parking position during parking. Acquiring the vehicle key point attitude data stream provides support for the subsequent generation of a standard vehicle key point attitude data stream.

[0039] The vehicle key point attitude data stream is filtered to generate a standard vehicle key point attitude data stream.

[0040] Data filtering refers to the process of digitizing a physical object, which inevitably introduces errors, redundancies, and measurement noise from the scanning environment. These errors significantly impact subsequent reconstruction of the physical model. To better extract the object's feature data, data filtering is necessary to remove these errors. In this application, it refers to filtering errors in the vehicle's key point poses to generate a standard vehicle key point pose data stream. This standard vehicle key point pose data stream refers to the pose data stream of a standard vehicle key point that has safely stopped without collisions or other anomalies during parking. By generating this standard vehicle key point pose data stream, the key point data stream of the standard vehicle's parking pose is obtained, laying the groundwork for setting pose change thresholds.

[0041] Determine the Kalman filter based on the filtering requirements parameters;

[0042] The vehicle key point attitude data stream is filtered based on the Kalman filter to generate the standard vehicle key point attitude data stream.

[0043] The filtering requirement parameters refer to the parameters of the data stream that need to be filtered. In this application, it refers to identifying the vehicle key point attitude data stream corresponding to the anomaly that occurred during parking, finding all abnormal vehicle key point attitude data streams, serializing them, and obtaining the minimum vehicle key point attitude data stream as the filtering requirement parameter. The Kalman filter is a highly efficient autoregressive filter that can estimate the state of a dynamic system in a combination of information with many uncertainties. It is a powerful and highly versatile tool for optimal estimation of the system state in dynamic systems containing uncertain information. The vehicle key point attitude data stream is filtered according to the Kalman filter to generate the standard vehicle key point attitude data stream. By obtaining the key point data stream of the standard vehicle parking attitude, support is provided for subsequent setting of attitude change thresholds.

[0044] Feature tracking analysis is performed on the standard vehicle key point attitude data stream to obtain key point attitude tracking feature information;

[0045] In the field of computer vision, feature tracking is a commonly used image analysis technique used to track targets or extract key features from images between consecutive frames. Feature tracking analysis is performed on the standard vehicle keypoint pose data stream to obtain keypoint pose tracking feature information, i.e., to obtain the feature information of the tracked target, specifically the pose information of key factors such as the vehicle body that significantly influence parking results. Obtaining this keypoint pose tracking feature information lays the groundwork for the subsequent keypoint pose tracking feature information judgment module.

[0046] like Figure 2 As shown, the standard vehicle key point attitude data stream is integrated according to the temporal changes to generate a vehicle key point attitude temporal data stream.

[0047] Obtain vehicle attitude factor information, including pitch angle, roll angle, and yaw angle;

[0048] Based on the vehicle attitude factor information, attitude calculation is performed on the time-series data stream of the vehicle key point attitude to determine the attitude tracking feature information of the key point.

[0049] Integrating the standard vehicle keypoint attitude data stream according to temporal changes refers to arranging the data in the order of its generation time to obtain a temporal data stream of vehicle keypoint attitude, acquiring vehicle attitude factor information, including pitch angle, roll angle, and yaw angle. Based on this vehicle attitude factor information, attitude calculation is performed on the temporal data stream of vehicle keypoint attitude. The Kalman filter is a more complex but more accurate attitude estimation method. It is based on state estimation and observation models, and combines measurement data with the system model through recursive processing. The Kalman filter considers measurement errors, system noise, and prior information, and optimizes the attitude estimation result by minimizing the mean square error. This method is very effective for high-precision attitude calculation, but requires more complex mathematical derivation and implementation. The keypoint attitude tracking feature information is then determined, where the keypoint attitude tracking feature information refers to specific keypoint attitude information. The keypoint attitude tracking feature information is determined by comparing it with the standard vehicle keypoint attitude data stream, laying the groundwork for subsequent judgment of the keypoint attitude tracking feature information.

[0050] Based on parking experience, a posture change threshold is set. When the posture tracking feature information of the key point exceeds the posture change threshold, the abnormal warning module is used to provide an abnormal warning for the posture during the parking process.

[0051] Parking experience refers to knowledge gained from multiple parking attempts. Based on this experience, a posture change threshold is set, where the posture change threshold is the critical value for a posture change—the minimum or maximum value at which an effect can occur. An anomaly warning module is set up to determine whether the parking posture is abnormal. The input data of this module is the key point posture tracking feature information, and the output data is the anomaly status of the key point posture tracking feature information and whether an anomaly warning is issued. When the key point posture tracking feature information exceeds the posture change threshold, the parking posture is abnormal; when the key point posture tracking feature information is less than the posture change threshold, the parking posture is not abnormal. Anomaly warnings are issued for example, when the parking azimuth angle is too large, it may encounter other vehicles or affect the parking position. By setting a posture change threshold and establishing an anomaly warning module to judge the key point posture tracking feature information, a rapid and accurate technical effect of judging parking anomalies is achieved, maintaining parking order.

[0052] When a parking anomaly warning is issued, the difference between the key point attitude tracking feature information and the attitude change threshold is used as the parking attitude anomaly parameter.

[0053] Based on the abnormal parking posture parameters, determine the parking posture correction parameters;

[0054] The parking posture of the target vehicle is corrected based on the parking posture correction parameters.

[0055] When a parking anomaly warning is issued, the difference between the key point attitude tracking feature information and the attitude change threshold is used as the parking attitude anomaly parameter. For example, if the steering angle is too large, the attitude parameters need to be corrected to ensure normal parking. Based on the parking attitude anomaly parameter, a parking attitude correction parameter is determined, where the parking attitude correction parameter refers to the parameter value of the feature that needs to be corrected. For example, if the parking steering angle is 100° and the steering angle threshold is 60°, then the parking attitude correction parameter is 40°. The parking attitude of the target vehicle is corrected based on the parking attitude correction parameter.

[0056] like Figure 3 As shown in the embodiment of this application, a key point attitude tracking and early warning system during berthing is also provided. The system includes:

[0057] Spatial distribution data information acquisition module 11, the spatial distribution data information acquisition module 11 is used to acquire spatial distribution data information of the target parking lot;

[0058] The berth space scene model generation module 12 is used to perform spatial modeling on the spatial distribution data information using three-dimensional modeling technology to generate a berth space scene model.

[0059] The target vehicle key point set acquisition module 13 is used to extract key points of the target vehicle based on the parking space scene model and acquire the target vehicle key point set.

[0060] Vehicle key point data stream acquisition module 14, the vehicle key point data stream acquisition module 14 is used to install an attitude sensor on the target vehicle, and to monitor the set of key points of the target vehicle in real time through the attitude sensor to acquire the vehicle key point attitude data stream.

[0061] Standard vehicle key point attitude data stream generation module 15 is used to perform data filtering on the vehicle key point attitude data stream to generate a standard vehicle key point attitude data stream.

[0062] The key point attitude tracking feature information acquisition module 16 is used to perform feature tracking analysis on the standard vehicle key point attitude data stream to obtain key point attitude tracking feature information.

[0063] An anomaly warning module 17 is used to set a posture change threshold based on parking experience. When the posture tracking feature information of the key point exceeds the posture change threshold, the anomaly warning module is used to provide an anomaly warning for the posture during the parking process.

[0064] Furthermore, embodiments of this application also include:

[0065] A parking lot construction design information acquisition module, which is used to acquire parking lot construction design information through a parking lot management system;

[0066] The parking space image information acquisition module is used to deploy a CCD image sensor in the target parking lot, and to collect images of the parking space occupancy status of the target parking lot through the CCD image sensor to obtain parking space image information.

[0067] A parking space occupancy vehicle feature information acquisition module is used to perform feature analysis on the parking space image information of the parking lot to obtain parking space occupancy vehicle feature information.

[0068] The spatial distribution data information determination module is used to determine the spatial distribution data information based on the parking lot construction and design information, the parking lot parking space image information, and the parking space occupancy vehicle characteristic information.

[0069] Furthermore, embodiments of this application also include:

[0070] The vehicle body color feature information determination module is used to construct a color space and perform color matching mapping based on the color space and the parking space image information to determine the vehicle body color feature information.

[0071] A structural dimension feature information acquisition module is used to perform structural extraction calculations on the parking space image information to acquire structural dimension feature information.

[0072] The module for obtaining vehicle occupancy feature information is used to fuse the structural size feature information and the vehicle body color feature to obtain the vehicle occupancy feature information of the parking space.

[0073] Furthermore, embodiments of this application also include:

[0074] A modeling data feature information acquisition module is used to acquire modeling data feature information, which includes feature type, spatial distribution location, and structural size.

[0075] A spatial distribution data feature information acquisition module is used to classify and label the spatial distribution data information based on the modeling data feature information to obtain spatial distribution data feature information;

[0076] The berth space scene model generation module is used to model the spatial distribution data information based on the spatial distribution data feature information using three-dimensional modeling technology to generate the berth space scene model.

[0077] Furthermore, embodiments of this application also include:

[0078] A vehicle key point attitude timing data stream generation module is used to integrate the standard vehicle key point attitude data stream according to the timing changes to generate a vehicle key point attitude timing data stream.

[0079] A vehicle attitude information acquisition module is used to acquire vehicle attitude factor information, including pitch angle, roll angle and yaw angle.

[0080] A key point attitude tracking feature information determination module is used to perform attitude calculation on the vehicle key point attitude time-series data stream based on the vehicle attitude factor information to determine the key point attitude tracking feature information.

[0081] Furthermore, embodiments of this application also include:

[0082] A Kalman filter determination module is used to determine the Kalman filter based on the filtering requirement parameters.

[0083] A vehicle key point attitude data stream generation module is used to perform data filtering on the vehicle key point attitude data stream based on the Kalman filter to generate the standard vehicle key point attitude data stream.

[0084] Furthermore, embodiments of this application also include:

[0085] A parking posture abnormality parameter acquisition module is used to obtain the difference between the key point posture tracking feature information and the posture change threshold as parking posture abnormality parameters when a parking abnormality warning is issued.

[0086] A parking posture correction parameter determination module is used to determine parking posture correction parameters based on the parking posture abnormality parameters.

[0087] A parking posture correction module is used to correct the parking posture of the target vehicle based on the parking posture correction parameters.

[0088] For specific embodiments of the key point attitude tracking and early warning system during berthing, please refer to the embodiments of the key point attitude tracking and early warning method during berthing described above, which will not be repeated here. The above modules can be embedded in hardware or independent of the processor in a computer device, or stored in software in the memory of a computer device, so that the processor can call and execute the operations corresponding to each module.

[0089] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0090] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for tracking and alerting key points of poses during a docking procedure, characterized in that, The method comprises: acquiring spatial distribution data information of a target parking lot; using three-dimensional modeling technology to perform spatial modeling on the spatial distribution data information to generate a parking space scene model; extracting key points of a target vehicle based on the parking space scene model to obtain a target vehicle key point set; installing a posture sensor on the target vehicle and monitoring the target vehicle key point set in real time through the posture sensor to obtain a vehicle key point posture data stream; performing data filtering on the vehicle key point posture data stream to generate a standard vehicle key point posture data stream; performing feature tracking analysis on the standard vehicle key point posture data stream to obtain key point posture tracking feature information; setting a posture change threshold value according to parking experience, and using an abnormality early warning module to perform abnormality early warning on the parking process posture when the key point posture tracking feature information exceeds the posture change threshold value.

2. The method of claim 1, wherein, The acquiring of the spatial distribution data information of the target parking lot comprises: acquiring parking lot construction design information through a parking lot management system; deploying a CCD image sensor in the target parking lot, collecting images of the parking space occupancy of the target parking lot through the CCD image sensor, and acquiring parking lot parking space image information; performing feature analysis on the parking lot parking space image information to obtain parking space occupied vehicle feature information; determining the spatial distribution data information based on the parking lot construction design information, the parking lot parking space image information, and the parking space occupied vehicle feature information.

3. The method of claim 2, wherein, The obtaining of the parking space occupied vehicle feature information comprises: constructing a color space, performing color matching mapping based on the color space and the parking lot parking space image information to determine vehicle body color feature information; performing structure extraction calculation on the parking lot parking space image information to obtain structure size feature information; fusing the structure size feature information and the vehicle body color feature to obtain the parking space occupied vehicle feature information.

4. The method of claim 1, wherein, The generation of the parking space scene model comprises: acquiring modeling data feature information, the modeling data feature information comprising feature object types, spatial distribution positions, and structure sizes; performing classification labeling on the spatial distribution data information based on the modeling data feature information to obtain spatial distribution data feature information; using three-dimensional modeling technology to model the spatial distribution data information based on the spatial distribution data feature information to generate the parking space scene model.

5. The method of claim 1, wherein, The obtaining of the key point posture tracking feature information comprises: integrating the standard vehicle key point posture data stream according to time sequence changes to generate a vehicle key point posture time sequence data stream; acquiring vehicle posture factor information, the vehicle posture factor information comprising a pitch angle, a roll angle, and a heading angle; performing posture calculation on the vehicle key point posture time sequence data stream based on the vehicle posture factor information to determine the key point posture tracking feature information.

6. The method of claim 1, wherein, The generation of the standard vehicle key point posture data stream comprises: determining a Kalman filter according to filtering requirement parameters; Filter the vehicle key point posture data stream based on the Kalman filter to generate the standard vehicle key point posture data stream.

7. The method of claim 1, wherein, The method comprises: When the parking abnormality early warning is performed, the difference between the key point posture tracking feature information and the posture change threshold is used as a parking posture abnormality parameter. According to the parking posture abnormality parameter, a parking posture correction parameter is determined. Based on the parking posture correction parameter, the target vehicle is corrected in parking posture.

8. A key point attitude tracking warning system in a berthing process, characterized by, The system comprises: A spatial distribution data information acquisition module is configured to acquire spatial distribution data information of a target parking lot. A parking space scene model generation module is configured to generate a parking space scene model by using a three-dimensional modeling technology to perform spatial modeling on the spatial distribution data information. A target vehicle key point set acquisition module is configured to extract key points of a target vehicle based on the parking space scene model to acquire a target vehicle key point set. A vehicle key point data stream acquisition module is configured to install a posture sensor on the target vehicle and acquire a vehicle key point posture data stream by monitoring the target vehicle key point set in real time through the posture sensor. A standard vehicle key point posture data stream generation module is configured to filter the vehicle key point posture data stream to generate a standard vehicle key point posture data stream. A key point posture tracking feature information acquisition module is configured to perform feature tracking analysis on the standard vehicle key point posture data stream to acquire key point posture tracking feature information. An abnormality early warning module is configured to set a posture change threshold according to parking experience and perform abnormality early warning on the posture during parking when the key point posture tracking feature information exceeds the posture change threshold.

Images