Parking lot point cloud feature extraction method and device, equipment and medium

Abstract

The embodiment of the present disclosure relates to a kind of parking lot point cloud feature extraction method, device, equipment and medium, wherein, method includes: obtaining the point cloud data and image of parking lot;From the point cloud data, the point cloud data of parking line is extracted, and the parking number is extracted from the image;According to the relative position relationship between the three-dimensional points contained in the point cloud data of parking line, the point cloud data of single parking line is separated from the point cloud data of parking line;According to the distance between parking number and single parking line in three-dimensional space, the parking number with the distance less than preset distance from single parking line is assigned to the point cloud data of single parking line.The scheme provided in the embodiment of the present disclosure can improve the distinguishability of parking line point cloud data, and provide accurate and reliable data basis for parking lot high-precision map mapping.

Description

Technical Field

[0001] This disclosure relates to the field of point cloud processing technology, and in particular to a method, apparatus, device and medium for extracting point cloud features from a parking lot. Background Technology

[0002] Currently, high-precision maps mainly focus on highways and some ordinary roads, lacking high-precision maps for parking lots. However, autonomous driving's last-mile automated parking requires high-precision maps. Furthermore, in the ride-hailing sector, pick-up points at train stations and airports are typically located in underground parking lots, and existing standard maps (SD maps) cannot meet the fast and accurate needs of drivers and passengers. In underground parking lots, due to GPS signal obstruction and height restrictions, existing high-precision map generation methods cannot be directly reused. Therefore, a point cloud feature extraction method for parking lots is urgently needed to meet the mapping requirements of high-precision parking lot maps. Summary of the Invention

[0003] To address the aforementioned technical problems, this disclosure provides a method, apparatus, device, and medium for extracting point cloud features from parking lots.

[0004] A first aspect of this disclosure provides a method for point cloud feature extraction from a parking lot. The method includes: acquiring point cloud data and an image of the parking lot; extracting point cloud data of parking space lines from the point cloud data and extracting parking space numbers from the image; separating point cloud data of a single parking space line from the point cloud data of the parking space lines based on the relative positional relationship between the three-dimensional points contained in the point cloud data of the parking space lines; and assigning parking space numbers whose distance from the single parking space line is less than a preset distance to the point cloud data of the single parking space line based on the parking space number and the distance of the single parking space line in three-dimensional space.

[0005] A second aspect of this disclosure provides a point cloud feature extraction apparatus, the apparatus comprising:

[0006] The acquisition module is used to acquire point cloud data and images of the parking lot;

[0007] The extraction module is used to extract the point cloud data of parking space lines from the point cloud data, and to extract the parking space number from the image.

[0008] The point cloud separation module is used to separate the point cloud data of a single parking line from the point cloud data of the parking line based on the relative positional relationship between the three-dimensional points contained in the point cloud data of the parking line.

[0009] The number assignment module is used to assign the parking space number to the point cloud data of a single parking space line if the distance between the parking space number and the single parking space line is less than a preset distance, based on the parking space number and the distance between the single parking space line and the single parking space line in three-dimensional space.

[0010] A third aspect of this disclosure provides a computer device including a memory and a processor, wherein the memory stores a computer program that, when executed by the processor, can implement the method of the first aspect described above.

[0011] A fourth aspect of this disclosure provides a computer-readable storage medium storing a computer program that, when executed by a computer device, causes the computer device to perform the method described in the first aspect.

[0012] A fifth aspect of this disclosure provides a computer program product stored in a storage medium, which, when executed by a processor of a computer device, causes the processor to perform the method described in the first aspect.

[0013] The technical solution provided in this disclosure has the following advantages compared with the prior art:

[0014] In this embodiment, after acquiring point cloud data and images of a parking lot, point cloud data of parking spaces is extracted from the point cloud data. Based on the relative positional relationship between the three-dimensional points contained in the point cloud data of the parking spaces, the point cloud data of a single parking space line can be accurately separated from the point cloud data. By identifying parking space numbers from the images, and based on the distance between the parking space number and the single parking space line in three-dimensional space, the parking space number that is less than a preset distance from the single parking space line is assigned to the point cloud data of the single parking space line. This accurately establishes the correspondence between the parking space number and the point cloud data of the parking space line, giving the point cloud data of the parking space line a number attribute. The number attribute can improve the distinguishability of the point cloud data of the parking space line. When the point cloud data of the parking space line with the number attribute is used as the data for the production of high-precision maps of parking lots, the accuracy of the high-precision mapping of parking lots can be improved. Attached Figure Description

[0015] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0016] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of a point cloud feature extraction scenario provided in an embodiment of this disclosure;

[0018] Figure 2 This is a flowchart of a point cloud feature extraction method for a parking lot provided in an embodiment of this disclosure;

[0019] Figure 3 This is a schematic diagram of an exemplary data partitioning method provided in an embodiment of this disclosure;

[0020] Figure 4 This is a flowchart of a method for extracting parking space line point cloud data according to an embodiment of this disclosure;

[0021] Figure 5 This is a schematic diagram of a point cloud classification method provided in an embodiment of this disclosure;

[0022] Figure 6 This is a schematic diagram of a method for extracting point cloud data of parking space lines provided in an embodiment of this disclosure;

[0023] Figure 7 This is a schematic diagram of a parking space line fitting method provided in an embodiment of this disclosure;

[0024] Figure 8 This is a schematic diagram of a parking space line fitting scenario provided by an embodiment of this disclosure;

[0025] Figure 9 This is a schematic diagram of the structure of a point cloud feature extraction device provided in an embodiment of this disclosure;

[0026] Figure 10 This is a schematic diagram of the structure of a computer device according to an embodiment of this disclosure. Detailed Implementation

[0027] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0028] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0029] In the process of creating high-precision maps of parking lots, the following methods are typically used to extract the features of parking lots:

[0030] Method 1 involves using computer vision algorithms to correct distortion in photos taken by the camera, then converting the distorted photos into orthophotos, and finally extracting the parking lines from the orthophotos. This method is affected by factors such as camera trajectory accuracy and the accuracy of image distortion parameter calculation, and therefore cannot meet high-precision requirements. Furthermore, image imaging is easily affected by lighting conditions, making it impossible to guarantee the accuracy of parking line extraction.

[0031] Method 2 involves using point cloud data detected by LiDAR in the parking lot to perform ground fitting processing to obtain ground point cloud data, and then extracting parking space line features from the ground point cloud data. However, the features of the parking space lines obtained in this method are too similar, resulting in weak distinguishability and hindering feature matching in the subsequent mapping process.

[0032] Method 3 involves converting the parking lot photo into an orthophoto after distortion correction, then using laser point cloud clustering to obtain the convex hulls of vehicles. These convex hulls are then projected onto the orthophoto image to remove interference from vehicles in the photo. Parking space information is then extracted from the photo. This method primarily utilizes the photo to extract lane lines, and its accuracy is still affected by the precision of distortion parameters and camera trajectory, resulting in relatively low accuracy. Furthermore, vehicle convex hull detection based on laser point clouds is prone to false detections, leading to the accidental deletion of photo information and impacting the accuracy of feature extraction.

[0033] To address the problems existing in related technologies and the mapping needs of parking lot scenarios, this disclosure provides a method for point cloud feature extraction in parking lots. For example, Figure 1 This is a schematic diagram of a point cloud feature extraction scenario provided in an embodiment of this disclosure, such as... Figure 1 As shown, this method extracts point cloud data of parking space lines from the point cloud data of a parking lot, extracts the parking space numbers contained in the parking lot from the parking lot image, and then, based on the relative positional relationship between the three-dimensional points contained in the point cloud data of the parking space lines, clusters adjacent three-dimensional points that are less than a preset clustering distance into a single parking space line, thereby obtaining the point cloud data of a single parking space line contained in the parking lot. Then, based on the parking space number and the distance of the single parking space line in three-dimensional space, the parking space number that is less than a preset distance from the single parking space line is assigned to the point cloud data of that parking space line, so that the point cloud data of the parking space line has a number attribute.

[0034] This method, after extracting the point cloud data of parking lines, can accurately separate the point cloud data of individual parking lines based on the relative positional relationships between 3D points. By assigning parking space numbers to the point cloud data of individual parking lines whose distances are less than a preset distance, the point cloud data of parking lines is endowed with a numbering attribute, which improves the distinguishability of the parking line point cloud data. In addition, extracting features of parking lines from point cloud data avoids the low accuracy problem of methods that extract parking lines based on photos.

[0035] It should be noted that, Figure 1 This is merely an exemplary implementation scenario, not the only one. For example, in other implementation scenarios, the extraction of parking space line point cloud data and the extraction of parking space numbers can also be performed sequentially. When executed sequentially, the extraction of parking space line point cloud data can occur before or after the extraction of parking space numbers.

[0036] To better understand the solutions of the embodiments of this disclosure, the solutions of the embodiments of this disclosure will be described below in conjunction with exemplary embodiments.

[0037] Figure 2 This is a flowchart illustrating a point cloud feature extraction method for a parking lot according to an embodiment of this disclosure. This method can be executed by a computer device, which can be understood as a device with computing and processing capabilities, such as a desktop computer, laptop computer, or server. Figure 2 As shown, the method includes:

[0038] Step 201: Obtain point cloud data and images of the parking lot.

[0039] Point cloud data can be understood as data collected by using a laser sensor (such as a lidar) mounted on a vehicle to emit a laser beam into a point in space and measuring the intensity (also known as reflectivity) of the reflected light and the time it takes for it to return. This allows the acquisition of spatial location information and material information (such as color, texture, and roughness) of that point. The collection of a large number of spatial points measured in this way is called laser point cloud data, or simply point cloud data.

[0040] Images of parking lots can be captured in the parking lot using camera equipment mounted on a vehicle (such as an RGB camera or a depth camera).

[0041] In one exemplary embodiment of this disclosure, a data acquisition vehicle can travel along a preset trajectory in a parking lot, collecting point cloud data and images of the surrounding environment during the journey. At each acquisition location (i.e., trajectory point), the acquisition vehicle simultaneously acquires point cloud data and images. In other words, the point cloud data and images in this embodiment are correlated in terms of acquisition location and acquisition time.

[0042] The point cloud data and images collected by the data acquisition vehicle can be associated and stored in a preset data source, such as a hard drive, database, or business server—devices with data storage capabilities. When executing the method of this embodiment, point cloud data and images of the parking lot can be obtained from the data source.

[0043] It should be noted that the point cloud data and images obtained from the data source in this embodiment can be the point cloud data and images of the entire parking lot, or the point cloud data and images of a portion of the parking lot. For example, when the parking lot is divided into multiple zones, the point cloud data and images of one zone can be obtained first, and then the feature extraction method of this embodiment can be used to extract features from the point cloud data of that zone. For example, if the parking lot includes four zones A, B, C, and D, the point cloud data and images of any one zone (e.g., zone A) can be obtained first for feature extraction, and then the point cloud data and images of another zone, such as zone B, can be obtained for feature extraction, until the point cloud feature extraction of the entire parking lot is completed.

[0044] In scenarios where point cloud data and images of the entire parking lot are acquired at once, the point cloud data and images can also be processed by regional partitioning. In this case, the point cloud data and images of the parking lot need to be divided, and then, based on the partitioning results, point cloud data and images of specific areas of the parking lot are acquired and processed. The basis for data partitioning can be the elevation fluctuation range of the acquisition trajectory. For example, point cloud data and images acquired on continuous trajectories with elevation fluctuations less than a preset threshold can be grouped into one segment, ensuring that the ground elevation corresponding to each segment is stable and ensuring the accuracy of feature extraction for each segment. Alternatively, in other implementations, the basis for data partitioning can be the parking lot's own regional division, dividing the point cloud data and images of each zone in the parking lot into one segment to ensure the continuity of parking space numbers in each segment. For example, if the parking lot includes four zones: A, B, C, and D, then the point cloud data and images of zone A can be divided into one segment, the point cloud data and images of zone B into another segment, the point cloud data and images of zone C into another segment, and the point cloud data and images of zone D into another segment.

[0045] Of course, the two data partitioning methods described above are merely exemplary and not the only methods. In fact, these two data partitioning methods can be combined in other implementations. For example, Figure 3 This is a schematic diagram of an exemplary data partitioning method provided in an embodiment of this disclosure. Figure 3In the method shown, after acquiring the point cloud data and images of the parking lot, the parking lot images can first be converted into orthographic images based on a preset transformation relationship. Then, the parking space numbers in the orthographic images are identified based on a preset character recognition method (such as Optical Character Recognition (OCR), but not limited to OCR). Since parking space numbers in parking lots generally have zoning information, for example, parking space numbers in zone A might be in the format A####, and parking space numbers in zone B might be in the format B####, the zoning information extracted from the parking space numbers can determine the parking space numbers belonging to the same zone. Then, based on the correspondence between the image and trajectory points, the parking space numbers are associated with the corresponding trajectory points, and based on the correspondence between the trajectory points and the point cloud data, the point cloud data belonging to the same zone is determined. Furthermore, based on the elevation fluctuation of trajectory points in the same partition, if the elevation fluctuation of the same partition is less than a preset threshold, no further division is required. If the elevation fluctuation of the same partition is greater than or equal to the preset threshold, the point cloud data and images in the same partition can be further divided according to the elevation fluctuation. The point cloud data and images in the sub-regions of the same partition with elevation fluctuation less than the preset threshold are divided into a segment.

[0046] exist Figure 3 In the data partitioning method shown, point cloud data and images of the same partition are grouped together by using the partition information in the parking space number. Then, based on the elevation change of the acquisition trajectory, point cloud data and images of areas with elevation fluctuations less than a preset threshold are grouped into a segment within the same partition. This not only ensures the stability of the ground elevation within the same segment but also guarantees the continuity of the parking space numbers within the same segment, providing data assurance for improving the accuracy of feature extraction.

[0047] Step 202: Extract the point cloud data of the parking space lines from the point cloud data, and extract the parking space number from the image.

[0048] This disclosure provides various methods for extracting parking space line point cloud data from point cloud data.

[0049] For example, in one exemplary method, point cloud data can be input into a preset first model, and the parking space line point cloud data can be extracted from the point cloud data through the first model. Here, the first model can be understood as a model with the ability to extract parking space line point clouds. This model can be trained using model training methods in related technologies, and the training process is not limited in this embodiment of the disclosure.

[0050] For example, in another exemplary method, point cloud data within a preset range around a preset ground reference height can be extracted from the point cloud data as ground point cloud data. Then, a preset second model is used to extract the parking line point cloud data from the ground point cloud data. By extracting ground point cloud data from the parking lot's point cloud data and then using the second model to extract the parking line point cloud data, the accuracy of parking line point cloud data extraction can be improved.

[0051] Of course, the two methods for extracting parking space line point cloud data mentioned above are just two exemplary methods, not the only methods.

[0052] For example, in this embodiment of the disclosure, there can be multiple methods for extracting parking space numbers. In one approach, parking space numbers can be extracted from a parking lot image using a preset number extraction model. In another approach, character recognition methods, such as OCR, can be used to identify the parking space numbers from the image. In practice, appropriate parking space number extraction methods can be selected as needed, and these will not be listed individually in this embodiment of the disclosure.

[0053] Step 203: Based on the relative positional relationship between the three-dimensional points contained in the point cloud data of the parking lines, separate the point cloud data of a single parking line from the point cloud data of the parking lines.

[0054] In practice, the distance between two adjacent 3D points on the same parking line is relatively close, while the distance between 3D points on different parking lines is relatively large. Based on this, in one embodiment of this disclosure, a clustering distance (hereinafter referred to as the second clustering distance for ease of distinction) is set. Then, according to the relative positional relationship between the 3D points contained in the point cloud data of the parking lines, adjacent 3D points whose distance to each other is less than the second clustering distance are clustered into the same parking line, thereby separating the point cloud data of a single parking line from the point cloud data of the parking lines.

[0055] In another embodiment of this disclosure, the distance between three-dimensional points can be determined based on the relative positional relationship between the three-dimensional points contained in the point cloud data of the parking lines, and the density between the three-dimensional points can be determined based on the mapping relationship between distance and density (the closer the distance, the higher the density). Then, based on the density between the three-dimensional points, a densely distributed area of ​​three-dimensional points can be identified in the point cloud data of the parking lines, and the point cloud data in a single densely distributed area can be identified as the point cloud data of a single parking line.

[0056] Step 204: Based on the parking space number and the distance of a single parking line in three-dimensional space, assign the parking space number that is less than a preset distance from the single parking line to the point cloud data of that parking line.

[0057] In practice, a parking space consists of four parking lines. For example, these four parking lines may include two longer lines in the vertical direction and two shorter lines in the horizontal direction. The single parking line referred to in this disclosure can be understood as one of the longer or shorter lines in the aforementioned parking space.

[0058] In this embodiment of the disclosure, the distance relationship between the parking space number and the parking space line (including the long side line in the vertical direction and / or the short side line in the horizontal direction) in the same parking space is statistically analyzed in advance, and then a distance threshold (i.e., preset distance) between the parking space number and the long side line and / or the short side line in the parking space is set according to the statistical results.

[0059] In one implementation, after obtaining the point cloud data of a single parking space line and the parking space number in the parking lot based on the methods of steps 202 and 203, the parking space number can be projected into the three-dimensional space to obtain the position of the parking space number in the three-dimensional space according to the mapping relationship between the image coordinate system of the extracted image of the parking space number and the three-dimensional coordinate system of the preset three-dimensional space (such as the three-dimensional space where the parking lot is located). Then, according to the parking space number and the distance of the single parking space line in the three-dimensional space, the parking space number that is less than the preset distance from the single parking space line is assigned to the point cloud data of the parking space line, so that the point cloud data of the parking space line has the number attribute.

[0060] In another implementation, for a given parking space line, if there are no parking space numbers within a preset distance of it, the nearest parking space line with a parking space number can be found from both sides of the parking space line (e.g., the left and right sides, or the top and bottom sides). Then, interpolation is performed on the numbers of these two found parking space lines to obtain an interpolated number, which is then assigned to the parking space line. For example, if the nearest parking space number to the left of a parking space line C is A123, and the nearest parking space number to the right of it is A125, the interpolation result is A124, which is then assigned to parking space line C.

[0061] When there are no parking space numbers less than a preset distance around a single parking space line, the problem of assigning parking space numbers when the parking space number is obscured can be solved by interpolating the numbers of the two parking space lines closest to it on both sides and assigning the interpolated number to the parking space line, thus improving the accuracy of parking space line point cloud data number assignment.

[0062] In this embodiment, after acquiring point cloud data and images of a parking lot, point cloud data of parking spaces is extracted from the point cloud data. Based on the relative positional relationship between the three-dimensional points contained in the point cloud data of the parking spaces, the point cloud data of a single parking space line can be accurately separated from the point cloud data. By identifying parking space numbers from the images, and based on the distance between the parking space number and the single parking space line in three-dimensional space, the parking space number that is less than a preset distance from the single parking space line is assigned to the point cloud data of the single parking space line. This accurately establishes the correspondence between the parking space number and the point cloud data of the parking space line, giving the point cloud data of the parking space line a number attribute. The number attribute can improve the distinguishability of the point cloud data of the parking space line. When the point cloud data of the parking space line with the number attribute is used as the data for the production of high-precision maps of parking lots, the accuracy of the high-precision mapping of parking lots can be improved.

[0063] Figure 4 This is a flowchart of a method for extracting parking space line point cloud data according to an embodiment of this disclosure, such as... Figure 4 As shown, the method includes:

[0064] Step 401: Classify the point cloud data to obtain ground point cloud data and facade point cloud data.

[0065] In this embodiment, the point cloud data can be understood as point cloud data of a portion of the parking lot, where the elevation fluctuation is less than a preset threshold and / or the area belongs to the same zone of the parking lot.

[0066] The facade point cloud data mentioned in this disclosure embodiment can be understood as point cloud data of a plane standing on the ground.

[0067] In this embodiment of the disclosure, the classification method for point cloud data may include various methods. For ease of understanding, several exemplary classification methods are illustrated below:

[0068] Method 1: Ground point cloud data and facade point cloud data can be extracted from parking lot point cloud data using a preset classification model.

[0069] Method 2: Plane fitting is used to fit the point cloud data of the parking lot. The plane with a fitted area greater than the preset area and a normal direction pointing upwards is defined as the ground. The plane with a normal direction pointing horizontally is defined as the elevation. Thus, the point cloud data on the ground is the ground point cloud data, and the point cloud data on the elevation is the elevation point cloud data.

[0070] Method 3 involves segmenting point cloud data into multiple voxels. A voxel is short for volume element. A solid containing voxels can be represented through stereo rendering or by extracting polygonal isosurfaces from a given threshold contour. A voxel is the smallest unit of 3D spatial segmentation. Principal Component Analysis (PCA) is performed on the voxels to identify the three directions in which point cloud distribution is most abundant. Then, based on the distribution fluctuations of the point cloud data in these three directions, planar voxels are extracted from the multiple voxels. A planar voxel is defined as a voxel whose fluctuation amplitude in one direction is greater than or equal to the absolute value of the difference between the fluctuation amplitudes in the other two directions and a preset amplitude. Further, within the planar voxels, the direction with the smallest fluctuation amplitude, i.e., the normal direction, is used to identify planar voxels pointing upwards towards the ground as ground voxels, and the direction with the smallest fluctuation amplitude pointing horizontally as elevation voxels. Thus, the point cloud data in ground voxels is ground point cloud data, and the point cloud data in elevation voxels is elevation point cloud data.

[0071] For example, Figure 5 This is a schematic diagram of a point cloud classification method provided in an embodiment of this disclosure, as shown below. Figure 5As shown, before classifying point cloud data, a trajectory point can be determined as a reference trajectory point from the acquisition trajectory of the point cloud data. When determining the reference trajectory point, a point can be arbitrarily selected from the acquisition trajectory corresponding to the point cloud data, or a trajectory point conforming to a preset rule can be selected as the reference trajectory point. For example, in one exemplary embodiment, a polygonal region containing the acquisition trajectory can be determined based on the position of the acquisition trajectory corresponding to the point cloud data. This polygonal region can be understood as the smallest polygonal region containing the acquisition trajectory, but is not limited to the smallest polygonal region. Further, the trajectory point on the acquisition trajectory closest to the center point of this polygonal region can be determined as the reference trajectory point. The coordinate system of the acquisition device (e.g., LiDAR) itself on the reference trajectory point is used as the reference coordinate system. By converting all point cloud data to the reference coordinate system, the interference of ground undulations on the identification of ground point cloud data and facade point cloud data is reduced. Specifically, for the point cloud data after coordinate conversion, this embodiment deletes point cloud data higher than the acquisition device based on the height of the acquisition device to reduce the amount of point cloud data. For the remaining point cloud data, the distribution space of the remaining point cloud data can be divided into multiple voxels by a preset voxel size. Each voxel can be understood as a cube of a preset size, containing the point cloud data divided by that cube. Within each voxel, the three directions with the most point cloud distribution can be determined using PCA, for example, denoted as the x-direction, y-direction, and z-direction. Assuming that the absolute value of the difference between the fluctuation amplitude in the x-direction and the fluctuation amplitude in the y-direction of the point cloud distribution in the voxel is greater than or equal to a preset amplitude, and the absolute value of the difference between the fluctuation amplitude in the x-direction and the fluctuation amplitude in the z-direction is also greater than or equal to a preset amplitude, then the voxel is determined to be a planar voxel. Further, if the direction with the smallest fluctuation amplitude, x, points upwards towards the ground, then the planar voxel is determined to be a ground voxel; if the direction x points horizontally, then the voxel is determined to be a vertical voxel. The point cloud data in the ground voxel is the ground point cloud data, and the point cloud data in the vertical voxel is the vertical point cloud data.

[0072] By dividing point cloud data into multiple voxels and then classifying ground point cloud data and elevation point cloud data according to the point cloud distribution in each voxel, the granularity of point cloud data classification is reduced and the accuracy of point cloud data classification is improved.

[0073] Step 402: Based on the reflection intensity information of the three-dimensional points contained in the ground point cloud data, extract the three-dimensional points with reflection intensity greater than the threshold intensity from the ground point cloud data as the point cloud data of the parking space line.

[0074] Example, Figure 6 This is a schematic diagram illustrating a method for extracting point cloud data of parking space lines according to an embodiment of this disclosure, as shown below. Figure 6 As shown, in one embodiment of this disclosure, a ground reflection intensity histogram can be established based on the reflection intensity information of three-dimensional points contained in the ground point cloud data. The horizontal axis of the histogram represents the reflection intensity, and the vertical axis represents the number of point clouds. Then, based on the ground reflection intensity histogram, the maximum inter-class variance method (OSTU) is used to determine the threshold intensity as described in this disclosure embodiment. The point cloud data with reflection intensity greater than the threshold intensity in the ground point cloud data is determined as the point cloud data of parking lines. The specific method for determining the threshold intensity using the maximum inter-class variance method can be found in related technologies and will not be elaborated here.

[0075] By determining the ground reflection intensity histogram and using the maximum inter-class variance method to determine the threshold intensity for judging parking space line point cloud data, the specificity of the threshold intensity for the current parking lot scenario can be improved, thereby improving the accuracy of parking space line point cloud data recognition.

[0076] For example, in one embodiment of this disclosure, after extracting the point cloud data of a single parking space line from the ground point cloud data and assigning a parking space number to the point cloud data of the single parking space line, the parking space number corresponding to the single parking space line closest to the facade point cloud data can be assigned to the facade point cloud data according to the distance between the facade point cloud data and the single parking space line.

[0077] By assigning parking space numbers to facade point cloud data, the distinguishability of the point cloud data can be further increased, and the information of the facades near the parking spaces can be clearly marked, providing data support for feature matching in the map building process.

[0078] For example, in some embodiments of this disclosure, after using three-dimensional points with reflection intensity greater than a threshold intensity in the ground point cloud data as the point cloud data for parking lines, the step of grouping the point cloud data for parking lines may be further included. Specifically, this includes:

[0079] S11. Based on the preset first clustering distance, perform Euclidean clustering on the point cloud data of parking lines to obtain at least one parking line point cloud group.

[0080] Considering that in actual parking lots, a row typically includes multiple consecutive parking spaces, and the parking space numbers of these consecutive spaces are usually also consecutive, this disclosure aims to group the parking space line point cloud data of multiple consecutive parking spaces in the same row into the same parking space line point cloud group to improve the accuracy of point cloud feature extraction. This embodiment sets a first clustering distance, enabling Euclidean clustering of the parking space line point cloud data based on this first clustering distance, so that the parking space line point cloud data of multiple consecutive parking spaces in the same row can be clustered into the same parking space line point cloud group. Specifically, the environment surrounding each 3D point in the parking space line point cloud data can be judged based on the first clustering distance. 3D points within the first clustering distance around each 3D point can be grouped into the same group. Then, groups with overlapping parts are merged together to form a single group, ultimately resulting in a parking space line point cloud group of multiple consecutively arranged parking spaces.

[0081] S12. Delete parking space line point cloud groups with a number of point clouds that is less than the preset number.

[0082] By clustering the point cloud data of continuously arranged parking spaces into a parking space point cloud group, and deleting parking space point cloud groups with fewer than a preset number of point clouds, misidentified parking space point cloud groups can be removed. Then, based on the relative positional relationship between the three-dimensional points contained in the remaining parking space point cloud groups, the point cloud data of a single parking space line contained in the parking space point cloud group can be accurately identified.

[0083] For example, in some further embodiments of this disclosure, before assigning parking space numbers less than a preset distance from a single parking space line to the point cloud data of that parking space line based on the parking space number and the distance of the single parking space line in three-dimensional space, a step of fitting the parking space line may be included. For example, Figure 7 This is a schematic diagram of a parking space line fitting method provided in an embodiment of this disclosure. Figure 7 As shown, the method includes:

[0084] S21. In the parking space line point cloud group, determine the unit parking space width based on the distance between the two parking space lines that are closest to the parking space number on both sides of the parking space number.

[0085] S22. Based on the distance between the two parking lines that are furthest from the parking space number on both sides, determine the total width of the multiple parking spaces included in the parking line point cloud group.

[0086] S23. Based on the total width of multiple parking spaces and the width of a single parking space, determine the number of parking spaces and the number of parking lines included in the parking line point cloud group.

[0087] S24. In response to the fact that the number of individual parking lines separated from the parking line point cloud group is less than the number of parking lines, fit parking lines in the parking line point cloud group so that the distance from all parking lines in the parking line point cloud group to the fitted parking lines is the shortest.

[0088] For example, Figure 8 This is a schematic diagram of a parking space line fitting scenario provided by an embodiment of this disclosure, such as... Figure 8 As shown, parking lines L1 and L2 are the two parking lines closest to parking space number "A####" on both sides. The distance h1 between L1 and L2 is the width per unit parking space. L3 and L4 are the two parking lines farthest from parking space number "A####" on both sides. The distance h2 between L3 and L4 is the total width of all parking spaces included in the parking line point cloud group. Assuming h2 is four times h1, then... Figure 8 It includes 4 parking spaces, which should correspond to 5 vertical parking lines. However, in reality... Figure 8 The system includes four parking space lines, but one is missing. This could be due to an object obstructing the view, preventing the capture of that parking space line. To resolve missing parking space lines caused by obstruction, you can... Figure 8 The parking space lines are fitted in the point cloud group shown, so that... Figure 8 The four parking lines in the diagram are the closest to the given parking line, and thus the result is obtained. Figure 8 Parking space lines are indicated by dashed lines.

[0089] By fitting parking lines when they are missing and filling in the missing lines, the integrity of the parking line point cloud data can be ensured, thus improving the accuracy of high-precision map construction.

[0090] Figure 9 This is a schematic diagram of a point cloud feature extraction device provided in an embodiment of this disclosure. This device can be exemplarily understood as the aforementioned computer device or a portion of a functional module within a computer device. For example... Figure 9 As shown, the point cloud feature extraction device 90 includes:

[0091] The acquisition module 91 is used to acquire point cloud data and images of the parking lot;

[0092] Extraction module 92 is used to extract point cloud data of parking lines from the point cloud data and to extract parking space numbers from the image;

[0093] The point cloud separation module 93 is used to separate the point cloud data of a single parking line from the point cloud data of the parking line based on the relative positional relationship between the three-dimensional points contained in the point cloud data of the parking line.

[0094] The numbering module 94 is used to assign the parking space number that is less than a preset distance from the single parking space line to the point cloud data of the single parking space line based on the parking space number and the distance between the single parking space line and the single parking space line in three-dimensional space.

[0095] In one embodiment, the acquisition module 91 can be used to acquire point cloud data and images of a portion of the parking lot, wherein the elevation fluctuation of the portion of the area is less than a preset threshold and / or the portion of the area belongs to the same zone of the parking lot.

[0096] In one implementation, the extraction module 92 may include:

[0097] The classification submodule is used to classify the point cloud data to obtain ground point cloud data and facade point cloud data.

[0098] The extraction submodule is used to extract three-dimensional points with a reflection intensity greater than a threshold intensity from the ground point cloud data based on the reflection intensity information of the three-dimensional points contained in the ground point cloud data, and use these points as the point cloud data for parking lines.

[0099] In one implementation, the numbering module 94 is also used to assign the parking space number corresponding to the single parking space line that is closest to the facade point cloud data to the facade point cloud data.

[0100] In one embodiment, the point cloud feature extraction device 90 may further include: a first processing module, configured to determine a trajectory point as a reference trajectory point from the acquisition trajectory of the point cloud data; use the coordinate system of the acquisition device on the reference trajectory point as a reference coordinate system, and convert the point cloud data to the reference coordinate system to obtain converted point cloud data; and delete point cloud data whose elevation is higher than that of the acquisition device after conversion to obtain the remaining point cloud data.

[0101] In one implementation, the first processing module is specifically used to: determine a polygonal region containing the acquisition trajectory based on the acquisition trajectory of the point cloud data; and determine the trajectory point closest to the center point of the polygonal region from the acquisition trajectory as a reference trajectory point.

[0102] In one implementation, the classification submodule is specifically used to: divide the remaining point cloud data into multiple voxels based on a preset voxel size;

[0103] For each voxel, based on the point cloud distribution in the voxel, the three directions with the most point cloud distribution in the voxel are determined; according to the distribution fluctuation of the point cloud data in the voxel in the three directions, planar voxels are extracted from the multiple voxels, wherein the planar voxel is a voxel whose absolute value of the difference between the fluctuation amplitude in one direction and the fluctuation amplitude in the other two directions is greater than or equal to a preset amplitude; planar voxels with the normal direction facing upwards are determined as ground voxels, and planar voxels with the normal direction facing horizontally are determined as elevation voxels; the point cloud data in the ground voxels are determined as ground point cloud data, and the point cloud data in the elevation voxels are determined as elevation point cloud data.

[0104] In one embodiment, the point cloud feature extraction device 90 may further include:

[0105] The generation module is used to generate a ground reflection intensity histogram based on the reflection intensity information of the three-dimensional points contained in the ground point cloud data.

[0106] The determination module is used to determine the threshold intensity based on the ground reflection intensity histogram using the maximum variance method.

[0107] In one embodiment, the point cloud feature extraction device 90 may further include: a second processing module, used to perform Euclidean clustering on the point cloud data of the parking lines based on a preset first clustering distance to obtain at least one parking line point cloud group; and delete parking line point cloud groups with a number of points less than a preset number.

[0108] The point cloud separation module 93 is used to cluster adjacent three-dimensional points that are less than the second clustering distance into the same parking line according to the relative positional relationship between the three-dimensional points contained in the remaining parking line point cloud group, so as to obtain the point cloud data of a single parking line contained in the parking line point cloud group.

[0109] In one embodiment, the point cloud feature extraction device 90 may further include: a fitting module, used for:

[0110] In the parking space line point cloud group, the unit parking space width is determined based on the distance between the two parking space lines that are closest to the parking space number on both sides of the parking space number.

[0111] Based on the distance between the two parking lines that are furthest from the parking space number on both sides of the parking space number, the total width of the multiple parking spaces included in the parking line point cloud group is determined.

[0112] Based on the total width of the multiple parking spaces and the width of the unit parking space, determine the number of parking spaces and the number of parking lines included in the parking line point cloud group;

[0113] In response to the fact that the number of individual parking lines separated from the parking line point cloud group is less than the total number of parking lines, parking lines are fitted in the parking line point cloud group such that all parking lines in the parking line point cloud group are closest to the fitted parking lines.

[0114] In one embodiment, the numbering module 94 is further configured to obtain the parking space numbers of the parking spaces that are closest to the single parking space line on both sides of the single parking space line and have parking space numbers; perform linear interpolation processing based on the obtained parking space numbers to obtain interpolated numbers; and assign the interpolated numbers to the point cloud data of the single parking space line.

[0115] The apparatus provided in this disclosure can execute any of the above-described method embodiments, and its execution method and beneficial effects are similar, so they will not be described again here.

[0116] This disclosure also provides a computer device including a processor and a memory, wherein the memory stores a computer program that, when executed by the processor, can implement the method of any of the above method embodiments.

[0117] Example, Figure 10 This is a schematic diagram of the structure of a computer device according to an embodiment of this disclosure. See below for details. Figure 10 The diagram illustrates a structural schematic suitable for implementing the computer device 1400 in the embodiments of this disclosure. The computer device 1400 in the embodiments of this disclosure may include, but is not limited to, devices with computing and data processing capabilities such as laptops, tablets, desktop computers, and servers. Figure 10 The computer device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0118] like Figure 10 As shown, computer device 1400 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 1401, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 1402 or a program loaded from storage device 1408 into random access memory (RAM) 1403. The RAM 1403 also stores various programs and data required for the operation of computer device 1400. The processing unit 1401, ROM 1402, and RAM 1403 are interconnected via bus 1404. Input / output (I / O) interface 1405 is also connected to bus 1404.

[0119] Typically, the following devices can be connected to I / O interface 1405: input devices 1406 including, for example, a touchscreen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; output devices 1407 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1408 including, for example, magnetic tape, hard disk, etc.; and communication devices 1409. Communication device 1409 allows computer device 1400 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 10 A computer device 1400 with various devices is shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have instead.

[0120] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 1404, or installed from storage device 1408, or installed from ROM 1402. When the computer program is executed by processing device 1401, it performs the functions defined in the methods of embodiments of this disclosure.

[0121] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0122] The aforementioned computer-readable medium may be included in the aforementioned computer device; or it may exist independently and not assembled into the computer device.

[0123] The aforementioned computer-readable medium carries one or more programs that, when executed by the computer device, cause the computer device to: acquire point cloud data and images of a parking lot; extract point cloud data of parking space lines from the point cloud data and extract parking space numbers from the images; separate point cloud data of individual parking space lines from the point cloud data of parking space lines based on the relative positional relationships between the three-dimensional points contained in the point cloud data of parking space lines; and assign parking space numbers whose distance from the individual parking space lines is less than a preset distance to the point cloud data of the individual parking space lines based on the parking space numbers and the distance of the individual parking space lines in three-dimensional space.

[0124] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including but not limited to object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0125] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0126] The units described in the embodiments of this disclosure can be implemented in software or in hardware. The names of the units are not, in some cases, intended to limit the specific unit.

[0127] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0128] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0129] This disclosure also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can perform the above-described functions. Figures 2-8 The methods in any of the embodiments are similar in execution and beneficial effects, and will not be described again here.

[0130] This disclosure also provides a computer program product stored in a storage medium, which, when executed by a processor of a computer device, causes the processor to perform... Figures 2-8 The methods in any of the embodiments are similar in execution and beneficial effects, and will not be described again here.

[0131] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0132] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for point cloud feature extraction in parking lots, wherein, include: Acquire point cloud data and images of the parking lot; Based on the acquisition trajectory of the point cloud data, a polygonal region containing the acquisition trajectory is determined; From the collected trajectory, determine the trajectory point that is closest to the center point of the polygonal region as the reference trajectory point; The coordinate system of the acquisition device on the reference trajectory point is used as the reference coordinate system, and the point cloud data is converted to the reference coordinate system to obtain the converted point cloud data; The point cloud data whose elevation is higher than that of the acquisition device after conversion is deleted to obtain the remaining point cloud data; The parking space line point cloud data is extracted from the remaining point cloud data, and the parking space number is extracted from the image. Based on the relative positional relationship between the three-dimensional points contained in the point cloud data of the parking lines, the point cloud data of a single parking line is separated from the point cloud data of the parking lines. Based on the parking space number and the distance of the single parking line in three-dimensional space, the parking space numbers that are less than a preset distance from the single parking line are assigned to the point cloud data of the single parking line.

2. The method according to claim 1, wherein, The acquisition of point cloud data and images of the parking lot includes: Obtain point cloud data and images of a portion of the parking lot, wherein the elevation fluctuation of the portion of the parking lot is less than a preset threshold and / or the portion of the parking lot belongs to the same zone.

3. The method according to claim 1 or 2, wherein, The step of extracting the parking space line point cloud data from the remaining point cloud data includes: The remaining point cloud data is classified and processed to obtain ground point cloud data and facade point cloud data; Based on the reflection intensity information of the three-dimensional points contained in the ground point cloud data, three-dimensional points with reflection intensity greater than a threshold intensity are extracted from the ground point cloud data as the point cloud data of the parking space lines.

4. The method according to claim 3, wherein, After assigning the parking space numbers that are less than a preset distance from the single parking space line to the point cloud data of the single parking space line, the method further includes: The parking space number corresponding to the single parking space line that is closest to the facade point cloud data is assigned to the facade point cloud data.

5. The method according to claim 3, wherein, The remaining point cloud data is classified to obtain ground point cloud data and facade point cloud data, including: Based on the preset voxel size, the remaining point cloud data is divided into multiple voxels; For each voxel, based on the point cloud distribution in the voxel, determine the three directions in which the point cloud distribution in the voxel is most abundant; Based on the distribution fluctuation of point cloud data in the voxels in the three directions, planar voxels are extracted from the plurality of voxels. The planar voxel is a voxel in which the absolute value of the difference between the fluctuation amplitude in one direction and the fluctuation amplitude in the other two directions is greater than or equal to a preset amplitude. Planar voxels with their normal direction pointing upwards from the ground are defined as ground voxels, and planar voxels with their normal direction pointing horizontally are defined as elevation voxels. The point cloud data in the ground voxels is determined as ground point cloud data, and the point cloud data in the facade voxels is determined as facade point cloud data.

6. The method according to claim 3, wherein, Before extracting the point cloud data for parking space lines from the ground point cloud data based on the reflection intensity information of the three-dimensional points contained in the ground point cloud data, the method includes: Based on the reflection intensity information of the three-dimensional points contained in the ground point cloud data, a ground reflection intensity histogram is generated; The threshold intensity is determined using the Otsu's method based on the ground reflection intensity histogram.

7. The method according to claim 3, wherein, After extracting the three-dimensional points with reflection intensity greater than a threshold intensity from the ground point cloud data as the point cloud data for parking space lines, the method further includes: Based on a preset first clustering distance, Euclidean clustering is performed on the point cloud data of the parking lines to obtain at least one parking line point cloud group. Delete parking space line point cloud groups whose point cloud count is less than the preset number; The step of separating the point cloud data of a single parking space line from the point cloud data of the parking space lines based on the relative positional relationships between the three-dimensional points contained in the point cloud data of the parking space lines includes: Based on the relative positional relationship between the three-dimensional points contained in the remaining parking space line point cloud group, adjacent three-dimensional points whose distance to each other is less than the second clustering distance are clustered into the same parking space line to obtain the point cloud data of a single parking space line contained in the parking space line point cloud group.

8. The method according to claim 7, wherein, Before assigning parking space numbers whose distance from a single parking space line is less than a preset distance to the point cloud data of the single parking space line based on the parking space number and the distance of the single parking space line in three-dimensional space, the method further includes: In the parking space line point cloud group, the unit parking space width is determined based on the distance between the two parking space lines that are closest to the parking space number on both sides of the parking space number. Based on the distance between the two parking lines that are furthest from the parking space number on both sides of the parking space number, the total width of the multiple parking spaces included in the parking line point cloud group is determined. Based on the total width of the multiple parking spaces and the width of the unit parking space, determine the number of parking spaces and the number of parking lines included in the parking line point cloud group; In response to the fact that the number of individual parking lines separated from the parking line point cloud group is less than the total number of parking lines, parking lines are fitted in the parking line point cloud group such that all parking lines in the parking line point cloud group are closest to the fitted parking lines.

9. The method according to claim 1, wherein, If there are no parking spaces near the single parking line that are less than the preset distance from the single parking line, the method further includes: Obtain the parking space number of the parking space line that is closest to the single parking space line on both sides and has a parking space number; Linear interpolation is performed based on the obtained parking space number to obtain the interpolated number; The interpolation number is assigned to the point cloud data of the single parking space line.

10. A point cloud feature extraction device, wherein, include: The acquisition module is used to acquire point cloud data and images of the parking lot; Based on the acquisition trajectory of the point cloud data, a polygonal region containing the acquisition trajectory is determined; From the collected trajectory, determine the trajectory point that is closest to the center point of the polygonal region as the reference trajectory point; The coordinate system of the acquisition device on the reference trajectory point is used as the reference coordinate system, and the point cloud data is converted to the reference coordinate system to obtain the converted point cloud data; The point cloud data whose elevation is higher than that of the acquisition device after conversion is deleted to obtain the remaining point cloud data; The extraction module is used to extract the point cloud data of the parking lines from the remaining point cloud data, and to extract the parking space number from the image. The point cloud separation module is used to separate the point cloud data of a single parking line from the point cloud data of the parking line based on the relative positional relationship between the three-dimensional points contained in the point cloud data of the parking line. The numbering assignment module is used to assign the parking space number that is less than a preset distance from the single parking space line to the point cloud data of the single parking space line, based on the parking space number and the distance between the single parking space line and the single parking space line in three-dimensional space.

11. A computer device, characterized in that, include: A memory and a processor, wherein the memory stores a computer program that, when executed by the processor, implements the method as described in any one of claims 1-9.

12. A computer program product, wherein, The program product is stored in a storage medium, and when the program product is executed by a processor in a computer device, the processor performs the method as described in any one of claims 1-9.

Images