Vehicle-mounted laser point cloud data position accuracy detection method and related device

Abstract

The present invention discloses a method and related device for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points. The method comprises: performing data acquisition and processing based on a vehicle-mounted laser device to obtain vehicle-mounted laser point cloud data; performing noise reduction preprocessing on the vehicle-mounted laser point cloud data to obtain road surface point cloud data; performing marking point cloud segmentation processing on the road surface point cloud data based on a preset algorithm, and then extracting traffic marking feature points from the segmented marking point cloud data to obtain traffic marking feature point data; and performing position accuracy detection processing using the traffic marking feature point data based on a same-accuracy detection method to obtain data accuracy detection results. In embodiments of the present invention, automated accuracy detection is achieved.

Description

Technical Field

[0001] The present invention relates to the field of precision calculation technology, and in particular to a method and related device for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points. Background Art

[0002] With the development of autonomous driving technology, intelligent transportation systems, and urban management systems, accurate perception of road environments is becoming increasingly important. Vehicle-mounted mobile laser scanning technology is a high-precision road environment perception data. The high-density three-dimensional point cloud data it generates has the advantages of high accuracy, rich information, and little weather influence, providing a reliable data foundation for the instantiation and analysis of road environment elements. However, point cloud quality still faces the following problems: due to the influence of global positioning system positioning errors, inertial navigation unit attitude errors, scanner angle and distance measurement errors, multi-sensor synchronization and calibration errors, etc., there are decimeter or even meter-level non-rigid deformations between round trips and revisits of vehicle-mounted point clouds at different times. Many scholars have been studying how to improve the position accuracy of point clouds. However, in actual production, vehicle-mounted point clouds collected in the field often have nonlinear and non-systematic errors, which require timely detection of point cloud position accuracy and error correction.

[0003] At present, some scholars have studied the quality inspection technology of airborne LiDAR data. Xu Guohong first clarified the content of airborne LiDAR data quality inspection. Li Haolin et al. proposed an airborne LiDAR data vulnerability detection method. Liu Rundong et al. proposed an automatic inspection algorithm for airborne LiDAR point cloud strip overlap. Wang Qianghui discussed point cloud quality inspection methods in terms of point cloud coverage, point cloud noise, edge connection accuracy, etc. She Yi et al. studied the automated inspection technology of point cloud density, strip overlap, route curvature, altitude maintenance, flight attitude and other inspection items, as well as the interactive inspection technology of strip splicing error, point cloud accuracy, flight attitude and other inspection items; Zheng Yanchun et al. studied the method of visually inspecting the quality of point cloud classification through multi-perspective, orthophoto and stereo methods; existing research focuses on airborne point cloud data, and the position accuracy detection of point clouds still uses manual inspection; the collection method and collection equipment of vehicle-mounted point cloud and airborne point cloud data are very different. The position accuracy of airborne point clouds is easy to control, and inspection work is easy to implement; however, vehicle-mounted mobile scanning requires planning the roads in the survey area into different survey sections, and collecting data for the lanes in the survey sections one by one, so it is necessary to check multiple points of different batches of scan data; in addition, the following problems will inevitably be encountered during the data collection process: point cloud holes caused by road vehicles and road surface water, roads that are too long or inconvenient to turn around, resulting in the two-way lane data of the same road not being in the same scanning batch, and after point cloud fusion, the error between point clouds cannot be eliminated through control point correction; these problems require repeated collection of point cloud data, and the point cloud position accuracy also needs to be checked multiple times; in summary, point cloud data position accuracy detection has problems such as large workload, high repeatability, and low degree of automation. Summary of the Invention

[0004] The purpose of the present invention is to overcome the shortcomings of the existing technology. The present invention provides a method and related devices for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points, which realizes automated detection of accuracy, greatly reduces the workload during detection, and improves the accuracy of detection, realizing a wider range of applications.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a method for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points, the method comprising:

[0006] Carry out data collection and processing based on vehicle-mounted laser equipment to obtain vehicle-mounted laser point cloud data;

[0007] Performing noise reduction preprocessing on the vehicle-mounted laser point cloud data to obtain road surface point cloud data;

[0008] Performing traffic marking point cloud segmentation processing on the road surface point cloud data based on a preset algorithm, and performing traffic marking feature point extraction processing on the segmented traffic marking point cloud data to obtain traffic marking feature point data;

[0009] Based on the same precision detection method, the traffic marking feature point data is used to perform position precision detection processing to obtain a data precision detection result.

[0010] Optionally, performing noise reduction preprocessing on the vehicle-mounted laser point cloud data to obtain road surface point cloud data includes:

[0011] performing a denoising process on the vehicle-mounted laser point cloud data by filtering out non-ground point cloud data based on a cloth simulation filtering algorithm to obtain vehicle-mounted laser point cloud data after a first denoising, wherein the cloth simulation filtering algorithm is an algorithm set using preset cloth rigidity parameters and preset cloth grid resolution parameters;

[0012] The road surface point cloud data of the vehicle-mounted laser point cloud data after the first noise reduction is extracted and processed based on the scanning line algorithm to obtain the road surface point cloud data.

[0013] Optionally, the extracting and processing the road surface point cloud from the vehicle-mounted laser point cloud data after the first noise reduction based on the scan line algorithm to obtain the road surface point cloud data includes:

[0014] Based on the scanning line algorithm, the horizontal distance and slope difference of the point cloud within a fixed window on each laser scanning line are used to find the road bump points. All the road bump points in the vehicle-mounted laser point cloud data after the first noise reduction are merged into the road bump points on the left and right sides.

[0015] The road bump points on the same side are sequentially connected to form a road edge line, and the road edge line is used to form a road contour. The road contour is then used to crop and extract the ground point cloud in the vehicle-mounted laser point cloud data after the first noise reduction to obtain road surface point cloud data.

[0016] Optionally, performing line marking point cloud segmentation processing on the road surface point cloud data based on a preset algorithm includes:

[0017] Performing clustering processing on the road surface point cloud data based on a K-means clustering algorithm to obtain a first clustering result;

[0018] Dividing the road surface point cloud data into road traffic marking point cloud data and non-road traffic marking point cloud data based on the first clustering result;

[0019] The road traffic marking point cloud data is subjected to denoising and optimization processing, and the denoised and optimized road traffic marking point cloud data is segmented into an independent road marking point cloud object based on a Euclidean distance algorithm to form segmented road marking point cloud data.

[0020] Optionally, the Euclidean distance algorithm is used to segment the denoised and optimized road traffic marking point cloud data into an independent road marking point cloud object to form segmented road marking point cloud data, including:

[0021] Initialize the point set Q so that the initialized point set Q is an empty set;

[0022] Randomly select any point in the denoised and optimized road traffic marking point cloud data as the initial point P;

[0023] Perform a proximity search on the initial point P to obtain all points within the set distance threshold r of the initial point P and add them to Q;

[0024] For all points in the point set Q that have not been searched for, a neighbor search is performed according to the set distance threshold r to obtain a cluster of line point cloud data within the set distance threshold r;

[0025] Until all points of the denoised and optimized road traffic marking point cloud data are traversed, segmented marking point cloud data is formed.

[0026] Optionally, the segmented traffic marking point cloud data is subjected to traffic marking feature point extraction processing to obtain traffic marking feature point data, including:

[0027] Set each segmentation mark point cloud data as set S, set the boundary point set E, and set the boundary point set E to be an empty set, and set the radius threshold to A;

[0028] Thus, if we randomly select two points B and C from the set S, and the distance between B and C is not greater than 2A, we generate two circumscribed circles with radius A that pass through both B and C.

[0029] If there are no other points in set S within any of the two circumscribed circles, then points B and C are added to the boundary point set E. This continues until all point pairs in set S are traversed, and the point cloud data in the boundary point set E is used as the outline boundary of the traffic marking.

[0030] The contour boundary is subjected to contour regularization processing, and traffic marking feature point data is obtained according to the contour regularization processing result.

[0031] Optionally, the same-precision detection method is used to perform position accuracy detection processing using the traffic marking feature point data to obtain a data accuracy detection result, including:

[0032] Selecting a preset number of corner points at intervals along the road direction as checkpoints, and obtaining real coordinate data of the checkpoints;

[0033] The point cloud coordinate data corresponding to the checkpoint is obtained based on the traffic marking feature point data, and the position accuracy detection processing is performed using the real coordinate data and the point cloud coordinate data of the checkpoint based on the same accuracy detection method to obtain the data accuracy detection result.

[0034] In addition, an embodiment of the present invention further provides a device for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points, the device comprising:

[0035] Data acquisition module: used to collect and process data based on vehicle-mounted laser equipment to obtain vehicle-mounted laser point cloud data;

[0036] Data denoising module: used for performing denoising preprocessing on the vehicle-mounted laser point cloud data to obtain road surface point cloud data;

[0037] Feature point extraction module: used to perform traffic marking point cloud segmentation processing on the road surface point cloud data based on a preset algorithm, and segment the traffic marking point cloud data to perform traffic marking feature point extraction processing to obtain traffic marking feature point data;

[0038] Accuracy detection module: used to perform position accuracy detection processing using the traffic marking feature point data based on the same accuracy detection method to obtain data accuracy detection results.

[0039] In addition, an embodiment of the present invention further provides an electronic device, including a processor and a memory, wherein the processor runs a computer program or code stored in the memory to implement the vehicle-mounted laser point cloud data position accuracy detection method as described in any one of the above.

[0040] In addition, an embodiment of the present invention further provides a computer-readable storage medium for storing a computer program or code. When the computer program or code is executed by a processor, the vehicle-mounted laser point cloud data position accuracy detection method as described in any one of the above is implemented.

[0041] In an embodiment of the present invention, data collection and processing are performed based on a vehicle-mounted laser device to obtain vehicle-mounted laser point cloud data; noise reduction pre-processing is performed on the vehicle-mounted laser point cloud data to obtain road surface point cloud data; marking point cloud segmentation processing is performed on the road surface point cloud data based on a preset algorithm, and the segmented marking point cloud data is subjected to traffic marking feature point extraction processing to obtain traffic marking feature point data; position accuracy detection processing is performed using the traffic marking feature point data based on the same accuracy detection method to obtain data accuracy detection results; automated accuracy detection is achieved, which greatly reduces the workload during detection, improves the accuracy of detection, and realizes a wider range of applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0043] Figure 1 1 is a flow chart of a method for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points in an embodiment of the present invention;

[0044] Figure 2 1 is a flow chart of a method for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points in another embodiment of the present invention;

[0045] Figure 3 Schematic diagram of the structure of a vehicle-mounted laser point cloud data position accuracy detection device based on road traffic marking feature points in an embodiment of the present invention;

[0046] Figure 4 is a schematic diagram of the structure of an electronic device in an embodiment of the present invention;

[0047] Figure 5 This is a schematic diagram of vehicle-mounted point cloud data for urban roads in an embodiment of the present invention;

[0048] Figure 6 is a schematic diagram of ground point cloud data in an embodiment of the present invention;

[0049] Figure 7 is a schematic diagram of road point cloud data in an embodiment of the present invention;

[0050] Figure 8 is a schematic diagram of the point cloud classification result in an embodiment of the present invention;

[0051] Figure 9 Schematic diagram of the outline of a road traffic marking in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0053] For example 1, please refer to Figure 1 , Figure 1 It is a flow chart of a method for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points in an embodiment of the present invention.

[0054] like Figure 1 As shown, a method for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points includes:

[0055] S101: Perform data acquisition and processing based on the vehicle-mounted laser equipment to obtain vehicle-mounted laser point cloud data;

[0056] In the specific implementation process of the present invention, a vehicle-mounted laser device is carried on the vehicle. When the vehicle is driving, the vehicle-mounted laser point cloud data can be obtained through data collection and processing performed by the vehicle-mounted laser device.

[0057] S102: performing noise reduction preprocessing on the vehicle-mounted laser point cloud data to obtain road surface point cloud data;

[0058] In a specific implementation of the present invention, the denoising preprocessing of the vehicle-mounted laser point cloud data to obtain road surface point cloud data includes: performing denoising processing on the vehicle-mounted laser point cloud data to filter out non-ground point cloud data in the vehicle-mounted laser point cloud data based on a cloth simulation filtering algorithm to obtain the vehicle-mounted laser point cloud data after the first denoising, wherein the cloth simulation filtering algorithm is an algorithm set using preset cloth rigidity parameters and preset cloth grid resolution parameters; and performing road surface point cloud extraction processing on the vehicle-mounted laser point cloud data after the first denoising based on a scan line algorithm to obtain road surface point cloud data.

[0059] Furthermore, the scanning line algorithm is used to extract and process the road surface point cloud from the vehicle-mounted laser point cloud data after the first denoising to obtain the road surface point cloud data, including: based on the scanning line algorithm, all the road bump points in the vehicle-mounted laser point cloud data after the first denoising are merged into the road bump points on the left and right sides by using the point cloud horizontal distance and slope difference within a fixed window on each laser scanning line to find the road bump points; the road bump points on the same side are sequentially connected into road sidelines, and the road sidelines are used to form a road contour, and the road contour is used to crop and extract the ground point cloud in the vehicle-mounted laser point cloud data after the first denoising to obtain the road surface point cloud data.

[0060] Specifically, since the on-board laser point cloud data collected by the on-board laser equipment has the characteristics of complex and diverse scenes, uneven point cloud density, and large amount of point cloud density data, the on-board point cloud contains a large number of non-ground points such as buildings, trees, and road facilities, and it is impossible to directly extract road traffic markings. Therefore, it is first necessary to filter and obtain ground points from the on-board point cloud; in the filtering process, there are morphological filtering algorithms based on opening operations, filtering algorithms based on triangulation, filtering algorithms that use slope differences to separate ground points and non-ground points, and cloth simulation filtering algorithms. Among them, the cloth simulation filtering algorithm has simple parameter settings and a wider range of applications, and has a good effect on filtering in dense urban artificial landforms; therefore, this embodiment adopts the cloth simulation filtering algorithm for execution, wherein the on-board laser point cloud data of urban roads is as follows Figure 5 shown.

[0061] Use the cloth simulation filtering algorithm to filter out non-ground points in the vehicle-mounted laser point cloud data. Since the terrain in urban scenes is usually flat, it is advisable to set a larger cloth rigidity parameter and a smaller cloth grid resolution parameter to ensure that buildings, structures, trees, etc. are effectively filtered out while better retaining road bumps. The ground point cloud after filtering is as follows: Figure 6 shown.

[0062] After filtering, the ground point cloud still retains a large number of non-road surface points, which affects the quality and speed of road marking extraction. Therefore, a scan line-based method is needed to further filter out road surface points. First, the road bump points are found based on the horizontal distance and slope difference of the point cloud within a fixed window on each laser scan line; then all road bump points are merged into two categories, and the road bump points on the same side are sequentially connected to form road edges; finally, the road contour composed of road edges is used to clip the ground point cloud to obtain the road surface point cloud. The road surface point cloud is as follows: Figure 7 shown.

[0063] That is, the scanning line algorithm is used to find the road bump points by using the horizontal distance and slope difference of the point cloud within a fixed window on each laser scanning line. All the road bump points in the vehicle-mounted laser point cloud data after the first denoising are merged into the road bump points on the left and right sides; and the road bump points on the same side are connected in sequence to form road sidelines, and the road sidelines are used to form a road contour. The road contour is then used to crop and extract the ground point cloud in the vehicle-mounted laser point cloud data after the first denoising to obtain road surface point cloud data.

[0064] S103: performing traffic marking point cloud segmentation processing on the road surface point cloud data based on a preset algorithm, and performing traffic marking feature point extraction processing on the segmented traffic marking point cloud data to obtain traffic marking feature point data;

[0065] In the specific implementation process of the present invention, the road surface point cloud data is subjected to marking point cloud segmentation processing based on a preset algorithm, including: clustering the road surface point cloud data based on a K-means clustering algorithm to obtain a first clustering result; dividing the road surface point cloud data into road traffic marking point cloud data and non-road traffic marking point cloud data based on the first clustering result; denoising and optimizing the road traffic marking point cloud data, and segmenting the denoised and optimized road traffic marking point cloud data into an independent road marking point cloud object based on a Euclidean distance algorithm to form segmented marking point cloud data.

[0066] Furthermore, the Euclidean distance algorithm divides the denoised and optimized road traffic marking point cloud data into an independent road marking point cloud object to form segmented marking point cloud data, including: initializing a point set Q so that the initialized point set Q is an empty set; randomly selecting any point in the denoised and optimized road traffic marking point cloud data as an initial point P; performing a neighbor search on the initial point P to obtain all points within a set distance threshold r of the initial point P and add them to Q; performing a neighbor search on all points in the point set Q that have not been subjected to a neighbor search according to the set distance threshold r to obtain a cluster of marking point cloud data within the set distance threshold r; until all points of the denoised and optimized road traffic marking point cloud data are traversed, segmented marking point cloud data is formed.

[0067] Furthermore, the segmented marking point cloud data is subjected to traffic marking feature point extraction processing to obtain traffic marking feature point data, including: setting each segmented marking point cloud data as a set S, and setting a boundary point set E, and the boundary point set E is an empty set, and setting the radius threshold to A; thereby arbitrarily selecting two points B and C in the set S, if the distance between point B and point C is not greater than 2A, then generating two circumscribed circles passing through point B and point C at the same time and with a radius of A; if there are no other points in the set S within any of the two circumscribed circles, then points B and C are added to the boundary point set E; until all point pairs in the set S are traversed, the point cloud data in the boundary point set E is used as the contour boundary of the traffic marking; performing contour regularization processing on the contour boundary, and obtaining traffic marking feature point data based on the contour regularization processing result.

[0068] Specifically, in the segmentation of road marking point clouds, the road surface point cloud is first divided into two categories: road traffic marking point clouds and non-road traffic marking point clouds based on the point cloud intensity information using the k-means clustering method. Then the road marking point cloud is denoised and optimized, and finally the Euclidean clustering method is used to segment the road marking point cloud into independent road marking point cloud objects.

[0069] For the classification of road surface point cloud data, this embodiment uses the k-means clustering method to separate the road marking point cloud from the road surface point cloud. This method is an unsupervised clustering algorithm that divides the sample space into k clusters based on the similarity between samples, so that the similarity within a cluster is high and the similarity between clusters is low. The road surface point cloud is divided into two clusters: road markings and non-road markings. The intensity of the laser point cloud reflected by road markings is significantly higher than that of non-road marking points. The similarity between clusters is calculated based on the intensity values ​​of all point clouds within the cluster. The specific implementation steps are as follows:

[0070] 1) Initial cluster center selection: select the point cloud sample m with the largest intensity and the point cloud sample r with the smallest intensity among all point clouds as the initial cluster center to avoid the influence of random selection of initial cluster centers on the classification results.

[0071] 2) Calculate the similarity distance between each point cloud sample x and the two initial cluster centers and , each point cloud sample is classified into the category of the cluster center closest to it, with M representing the road marking category and R representing the non-road marking category.

[0072] 3) Recalculate the cluster centers; calculate the new cluster centers of the two types of samples respectively , ,in represents the number of samples of category M, Indicates the number of samples of category R.

[0073] 4) Repeat steps 2) and 3) until the cluster center no longer changes.

[0074] 5) Normalization: Set the intensity values ​​of all point clouds in category M to 255, and set the intensity values ​​of all point clouds in category R to 0.

[0075] In terms of noise point removal, the clustered road traffic marking point cloud may contain some isolated points with abnormal intensity. Radius filtering can be used to remove isolated points. For each point cloud, the number of points within a specified radius (for example, 10 times the average point spacing of the point cloud) around the search point is searched. If the number of points is lower than a certain threshold, it is considered an isolated point; for example, Figure 8 shown.

[0076] In the Euclidean clustering segmentation, the classified road traffic marking point cloud is a whole, and each marking needs to be segmented into independent individuals. Due to the large distance between road markings, the classic clustering method Euclidean clustering method can be used for segmentation. The specific implementation steps are as follows:

[0077] 1) Initialize the point set Q = empty set and randomly select an initial point P in the point cloud.

[0078] 2) Perform a neighbor search on P to obtain all its points within the set distance threshold r and add them to the point set Q.

[0079] 3) Perform proximity search on all points in the point set Q that have not been searched for within the distance threshold r.

[0080] 4) Repeat steps 2) and 3) to obtain a cluster of point clouds that meet the distance threshold.

[0081] 5) Reset the initial point P of the unclustered point cloud and repeat the above steps until all point clouds are divided into different point cloud sets; that is, each point cloud set corresponds to a segmentation mark point cloud data.

[0082] When extracting road marking features, the contour of the segmented road traffic marking point cloud is first extracted and regularized, and the corner points of the contour are the marking feature points.

[0083] Contour extraction is the process of extracting contours from a set of spatial points. Common algorithms include the convex hull algorithm, the Delaunay triangulation method, algorithms based on nearest neighbor geometric features, and the Alpha-Shapes algorithm. The Alpha-Shapes algorithm was chosen because road markings are not necessarily convex hulls. This algorithm was first proposed by Edelsbrunner et al. in 1983 and has since been refined and applied to fields such as image processing and point cloud data processing. To improve processing efficiency, the 3D point cloud can be pre-projected onto a horizontal plane to convert it into a 2D point cloud before contour extraction. The specific implementation steps are as follows:

[0084] 1) Let the road marking point cloud set be S, the boundary point set E = empty set, and set the radius threshold A.

[0085] 2) Select any two points B and C from S.

[0086] 3) If the distance between points B and C is not greater than 2A, generate two circumcircles that pass through both points B and C and have a radius of A. If no other point in S exists within either of these circumcircles, add these two points to the set of points E.

[0087] 4) Select two more points B and C from S and repeat step 3) until all point pairs are traversed. The point set E is the contour boundary.

[0088] Contour regularization: Directly connecting the boundary points of the contour extraction will result in a contour that is too rough, so the contour needs to be regularized. The Douglas-Peuker algorithm (DP algorithm) was proposed by David H. Douglas and Thomas K. Peucker in 1973 and is mainly used to simplify curves or polylines. The specific implementation steps are as follows

[0089] 1) Connect the first and last points A and B of the broken line to generate a straight line L.

[0090] 2) Calculate the distances from all points on the polyline to the straight line L in sequence.

[0091] 3) Set a distance threshold t. If the maximum distance D among all distances is less than the distance threshold t, delete all points on the polyline except A and B; if D is greater than the threshold t, split the polyline into two segments from the maximum distance point.

[0092] 4) Repeat steps 2) and 3) for all polyline segments until all distances D are less than the threshold.

[0093] The result after contour regularization is as follows Figure 9As shown, the traffic marking feature point data can be extracted.

[0094] S104: performing position accuracy detection processing using the traffic marking feature point data based on the same accuracy detection method to obtain a data accuracy detection result.

[0095] In the specific implementation process of the present invention, the same-precision detection method uses the traffic marking feature point data to perform position accuracy detection processing to obtain data accuracy detection results, including: selecting a preset number of corner points as checkpoints at intervals along the road direction, and obtaining the real coordinate data of the checkpoints; obtaining the point cloud coordinate data corresponding to the checkpoint based on the traffic marking feature point data, and using the real coordinate data and point cloud coordinate data of the checkpoint based on the same-precision detection method to perform position accuracy detection processing to obtain data accuracy detection results.

[0096] Specifically, the same precision detection method is used to perform precision detection on the vehicle-mounted laser point cloud data. The corner points of the road markings are obvious feature points. One or two corner points are selected at a certain distance along the road direction as checkpoints. The true coordinate values ​​of the checkpoints can be collected using the field RTK method. The point cloud coordinate values ​​corresponding to the checkpoints can be obtained by extracting the feature points of the road traffic markings. With the checkpoint as the center, the road traffic marking feature points within a certain radius are searched, and the difference between the two is calculated to detect the point accuracy; the calculation formula is as follows:

[0097] ;

[0098] ;

[0099] in, represents the error in the plane of the checkpoint; Indicates the error in the elevation of the checkpoint; Indicates the total number of checkpoints; Represents the difference between the true value of check point i and the point cloud feature point value in the X direction; Indicates the difference between the true value of the check point i and the point cloud feature point value in the Y direction; Indicates the elevation value of the point cloud feature point; Indicates the true value of the elevation of the checkpoint.

[0100] Experiments and analyses were conducted on the above-mentioned precision detection. To verify the feasibility and efficiency of this precision detection method, vehicle-mounted laser point cloud data containing different road scenes within an area of ​​10 square kilometers were selected for testing. The Tianbao vehicle-mounted 3D laser scanner MX50 was used to obtain point cloud data of the main roads and secondary roads in the test area. A total of 44 mobile measurements were completed in the survey area, with a total route of approximately 151.13 kilometers and an effective route of approximately 129.58 kilometers.

[0101] In order to detect the position accuracy of the vehicle-mounted point cloud, a total of 148 checkpoints were selected, and the points were all selected as road traffic marking feature points. The method proposed in this paper was used for detection, and the point cloud feature points corresponding to all checkpoints could be extracted. By comparing the checkpoints and point cloud feature points, the point cloud accuracy could be accurately detected. The accuracy calculation results are shown in Table 1.

[0102] Table 1 Point cloud accuracy statistics

[0103]

[0104] Note: The first 4 digits of both X and Y coordinates are removed.

[0105] Compared with traditional accuracy detection methods, the vehicle-mounted laser point cloud data position accuracy detection method based on road traffic marking feature points has obvious efficiency advantages; if the traditional method is used, it is necessary to manually select the feature points corresponding to the checkpoints in the point cloud data, which is time-consuming and labor-intensive, and has a low degree of automation; one engineer needs to spend 48 hours to complete the accuracy detection of 129.58 kilometers of vehicle-mounted point clouds in the test area; if the new method is used, using a graphics workstation (Core i9 processor, 32G memory, NVIDIA Quadro RTX4000 graphics card) to process point cloud data, one engineer only needs 27 hours to complete the accuracy detection of vehicle-mounted point clouds in the test area, and all processing is done automatically by computer; if a more powerful computing platform is used, it is possible to jointly process point cloud data of multiple roads, further improving detection efficiency.

[0106] In an embodiment of the present invention, data collection and processing are performed based on a vehicle-mounted laser device to obtain vehicle-mounted laser point cloud data; noise reduction pre-processing is performed on the vehicle-mounted laser point cloud data to obtain road surface point cloud data; marking point cloud segmentation processing is performed on the road surface point cloud data based on a preset algorithm, and the segmented marking point cloud data is subjected to traffic marking feature point extraction processing to obtain traffic marking feature point data; position accuracy detection processing is performed using the traffic marking feature point data based on the same accuracy detection method to obtain data accuracy detection results; automated accuracy detection is achieved, which greatly reduces the workload during detection, improves the accuracy of detection, and realizes a wider range of applications.

[0107] For example 2, please refer to Figure 2 , Figure 2 It is a flow chart of a method for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points in another embodiment of the present invention.

[0108] like Figure 2 As shown, a method for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points includes:

[0109] S201: performing data acquisition and processing based on the vehicle-mounted laser equipment to obtain vehicle-mounted laser point cloud data;

[0110] S202: performing noise reduction preprocessing on the vehicle-mounted laser point cloud data to obtain road surface point cloud data;

[0111] S203: performing clustering processing on the road surface point cloud data based on a K-means clustering algorithm to obtain a first clustering result;

[0112] S204: dividing the road surface point cloud data into road traffic marking point cloud data and non-road traffic marking point cloud data based on the first clustering result;

[0113] S205: performing denoising and optimization processing on the road traffic marking point cloud data, and segmenting the denoised and optimized road traffic marking point cloud data into independent road marking point cloud objects based on a Euclidean distance algorithm to form segmented road marking point cloud data;

[0114] S206: setting each segmentation mark point cloud data as a set S, setting a boundary point set E, and setting the boundary point set E to be an empty set, and setting the radius threshold to A;

[0115] S207: Thus, two points B and C are randomly selected from the set S. If the distance between point B and point C is not greater than 2A, two circumscribed circles with a radius of A passing through both points B and C are generated.

[0116] S208: If there are no other points in the set S within any of the two circumscribed circles, then point B and point C are added to the boundary point set E. This continues until all point pairs in the set S are traversed, and the point cloud data in the boundary point set E is used as the outline boundary of the traffic marking.

[0117] S209: performing contour regularization processing on the contour boundary, and obtaining traffic marking feature point data according to the contour regularization processing result;

[0118] S210: Performing position accuracy detection processing using the traffic marking feature point data based on the same accuracy detection method to obtain a data accuracy detection result.

[0119] The specific implementation of Example 2 can be found in the above examples, which will not be described in detail here.

[0120] For example three, please refer to Figure 3 , Figure 3 The figure is a schematic diagram of the structural composition of a vehicle-mounted laser point cloud data position accuracy detection device based on road traffic marking feature points in an embodiment of the present invention.

[0121] like Figure 3As shown, a vehicle-mounted laser point cloud data position accuracy detection device based on road traffic marking feature points includes:

[0122] Data acquisition module 301: used to perform data acquisition and processing based on the vehicle-mounted laser equipment to obtain vehicle-mounted laser point cloud data;

[0123] In the specific implementation process of the present invention, a vehicle-mounted laser device is carried on the vehicle. When the vehicle is driving, the vehicle-mounted laser point cloud data can be obtained through data collection and processing performed by the vehicle-mounted laser device.

[0124] Data denoising module 302: used for performing denoising preprocessing on the vehicle-mounted laser point cloud data to obtain road surface point cloud data;

[0125] In a specific implementation of the present invention, the denoising preprocessing of the vehicle-mounted laser point cloud data to obtain road surface point cloud data includes: performing denoising processing on the vehicle-mounted laser point cloud data to filter out non-ground point cloud data in the vehicle-mounted laser point cloud data based on a cloth simulation filtering algorithm to obtain the vehicle-mounted laser point cloud data after the first denoising, wherein the cloth simulation filtering algorithm is an algorithm set using preset cloth rigidity parameters and preset cloth grid resolution parameters; and performing road surface point cloud extraction processing on the vehicle-mounted laser point cloud data after the first denoising based on a scan line algorithm to obtain road surface point cloud data.

[0126] Furthermore, the scanning line algorithm is used to extract and process the road surface point cloud from the vehicle-mounted laser point cloud data after the first denoising to obtain the road surface point cloud data, including: based on the scanning line algorithm, all the road bump points in the vehicle-mounted laser point cloud data after the first denoising are merged into the road bump points on the left and right sides by using the point cloud horizontal distance and slope difference within a fixed window on each laser scanning line to find the road bump points; the road bump points on the same side are sequentially connected into road sidelines, and the road sidelines are used to form a road contour, and the road contour is used to crop and extract the ground point cloud in the vehicle-mounted laser point cloud data after the first denoising to obtain the road surface point cloud data.

[0127] Specifically, since the on-board laser point cloud data collected by the on-board laser equipment has the characteristics of complex and diverse scenes, uneven point cloud density, and large amount of point cloud density data, the on-board point cloud contains a large number of non-ground points such as buildings, trees, and road facilities, and it is impossible to directly extract road traffic markings. Therefore, it is first necessary to filter and obtain ground points from the on-board point cloud; in the filtering process, there are morphological filtering algorithms based on opening operations, filtering algorithms based on triangulation, filtering algorithms that use slope differences to separate ground points and non-ground points, and cloth simulation filtering algorithms. Among them, the cloth simulation filtering algorithm has simple parameter settings and a wider range of applications, and has a good effect on filtering in dense urban artificial landforms; therefore, this embodiment adopts the cloth simulation filtering algorithm for execution, wherein the on-board laser point cloud data of urban roads is as follows Figure 5 shown.

[0128] Use the cloth simulation filtering algorithm to filter out non-ground points in the vehicle-mounted laser point cloud data. Since the terrain in urban scenes is usually flat, it is advisable to set a larger cloth rigidity parameter and a smaller cloth grid resolution parameter to ensure that buildings, structures, trees, etc. are effectively filtered out while better retaining road bumps. The ground point cloud after filtering is as follows: Figure 6 shown.

[0129] The filtered ground point cloud still retains a large number of non-road surface points, which affects the extraction quality and speed of road markings. Therefore, it is necessary to use a scan line-based method to further filter out road surface points. First, find the road bump points based on the horizontal distance and slope difference of the point cloud within the fixed window on each laser scan line; then merge all the road bump points into two categories, and connect the road bump points on the same side in sequence to form road sidelines; finally, use the road contour composed of road sidelines to clip the ground point cloud to obtain the road surface point cloud. The road surface point cloud is as follows: Figure 7 shown.

[0130] That is, the scanning line algorithm is used to find the road bump points by using the horizontal distance and slope difference of the point cloud within a fixed window on each laser scanning line. All the road bump points in the vehicle-mounted laser point cloud data after the first denoising are merged into the road bump points on the left and right sides; and the road bump points on the same side are connected in sequence to form road sidelines, and the road sidelines are used to form a road contour. The road contour is then used to crop and extract the ground point cloud in the vehicle-mounted laser point cloud data after the first denoising to obtain road surface point cloud data.

[0131] Feature point extraction module 303: used to perform traffic marking point cloud segmentation processing on the road surface point cloud data based on a preset algorithm, and segment the traffic marking point cloud data to perform traffic marking feature point extraction processing to obtain traffic marking feature point data;

[0132] In the specific implementation process of the present invention, the road surface point cloud data is subjected to marking point cloud segmentation processing based on a preset algorithm, including: clustering the road surface point cloud data based on a K-means clustering algorithm to obtain a first clustering result; dividing the road surface point cloud data into road traffic marking point cloud data and non-road traffic marking point cloud data based on the first clustering result; denoising and optimizing the road traffic marking point cloud data, and segmenting the denoised and optimized road traffic marking point cloud data into an independent road marking point cloud object based on a Euclidean distance algorithm to form segmented marking point cloud data.

[0133] Furthermore, the Euclidean distance algorithm divides the denoised and optimized road traffic marking point cloud data into an independent road marking point cloud object to form segmented marking point cloud data, including: initializing a point set Q so that the initialized point set Q is an empty set; randomly selecting any point in the denoised and optimized road traffic marking point cloud data as an initial point P; performing a neighbor search on the initial point P to obtain all points within a set distance threshold r of the initial point P and add them to Q; performing a neighbor search on all points in the point set Q that have not been subjected to a neighbor search according to the set distance threshold r to obtain a cluster of marking point cloud data within the set distance threshold r; until all points of the denoised and optimized road traffic marking point cloud data are traversed, segmented marking point cloud data is formed.

[0134] Furthermore, the segmented marking point cloud data is subjected to traffic marking feature point extraction processing to obtain traffic marking feature point data, including: setting each segmented marking point cloud data as a set S, and setting a boundary point set E, and the boundary point set E is an empty set, and setting the radius threshold to A; thereby arbitrarily selecting two points B and C in the set S, if the distance between point B and point C is not greater than 2A, then generating two circumscribed circles passing through point B and point C at the same time and with a radius of A; if there are no other points in the set S within any of the two circumscribed circles, then points B and C are added to the boundary point set E; until all point pairs in the set S are traversed, the point cloud data in the boundary point set E is used as the contour boundary of the traffic marking; performing contour regularization processing on the contour boundary, and obtaining traffic marking feature point data based on the contour regularization processing result.

[0135] Specifically, in the segmentation of road marking point clouds, the road surface point cloud is first divided into two categories: road traffic marking point clouds and non-road traffic marking point clouds based on the point cloud intensity information using the k-means clustering method. Then the road marking point cloud is denoised and optimized, and finally the Euclidean clustering method is used to segment the road marking point cloud into independent road marking point cloud objects.

[0136] For the classification of road surface point cloud data, this embodiment uses the k-means clustering method to separate the road marking point cloud from the road surface point cloud. This method is an unsupervised clustering algorithm that divides the sample space into k clusters based on the similarity between samples, so that the similarity within a cluster is high and the similarity between clusters is low. The road surface point cloud is divided into two clusters: road markings and non-road markings. The intensity of the laser point cloud reflected by road markings is significantly higher than that of non-road marking points. The similarity between clusters is calculated based on the intensity values ​​of all point clouds within the cluster. The specific implementation steps are as follows:

[0137] 1) Initial cluster center selection: select the point cloud sample m with the largest intensity and the point cloud sample r with the smallest intensity among all point clouds as the initial cluster center to avoid the influence of random selection of initial cluster centers on the classification results.

[0138] 2) Calculate the similarity distance between each point cloud sample x and the two initial cluster centers and , each point cloud sample is classified into the category of the cluster center closest to it, with M representing the road marking category and R representing the non-road marking category.

[0139] 3) Recalculate the cluster centers; calculate the new cluster centers of the two types of samples respectively , ,in represents the number of samples of category M, Indicates the number of samples of category R.

[0140] 4) Repeat steps 2) and 3) until the cluster center no longer changes.

[0141] 5) Normalization: Set the intensity values ​​of all point clouds in category M to 255, and set the intensity values ​​of all point clouds in category R to 0.

[0142] In terms of noise point removal, the clustered road traffic marking point cloud may contain some isolated points with abnormal intensity. Radius filtering can be used to remove isolated points. For each point cloud, the number of points within a specified radius (for example, 10 times the average point spacing of the point cloud) around the search point is searched. If the number of points is lower than a certain threshold, it is considered an isolated point; for example, Figure 8 shown.

[0143] In the Euclidean clustering segmentation, the classified road traffic marking point cloud is a whole, and each marking needs to be segmented into independent individuals. Due to the large distance between road markings, the classic clustering method Euclidean clustering method can be used for segmentation. The specific implementation steps are as follows:

[0144] 1) Initialize the point set Q = empty set and randomly select an initial point P in the point cloud.

[0145] 2) Perform a neighbor search on P to obtain all its points within the set distance threshold r and add them to the point set Q.

[0146] 3) Perform proximity search on all points in the point set Q that have not been searched for within the distance threshold r.

[0147] 4) Repeat steps 2) and 3) to obtain a cluster of point clouds that meet the distance threshold.

[0148] 5) Reset the initial point P of the unclustered point cloud and repeat the above steps until all point clouds are divided into different point cloud sets; that is, each point cloud set corresponds to a segmentation mark point cloud data.

[0149] When extracting road marking features, the contour of the segmented road traffic marking point cloud is first extracted and regularized, and the corner points of the contour are the marking feature points.

[0150] Contour extraction is the process of extracting contours from a set of spatial points. Common algorithms include the convex hull algorithm, the Delaunay triangulation method, algorithms based on nearest neighbor geometric features, and the Alpha-Shapes algorithm. The Alpha-Shapes algorithm was chosen because road markings are not necessarily convex hulls. This algorithm was first proposed by Edelsbrunner et al. in 1983 and has since been refined and applied to fields such as image processing and point cloud data processing. To improve processing efficiency, the 3D point cloud can be pre-projected onto a horizontal plane to convert it into a 2D point cloud before contour extraction. The specific implementation steps are as follows:

[0151] 1) Let the road marking point cloud set be S, the boundary point set E = empty set, and set the radius threshold A.

[0152] 2) Select any two points B and C from S.

[0153] 3) If the distance between points B and C is not greater than 2A, generate two circumcircles that pass through both points B and C and have a radius of A. If no other point in S exists within either of these circumcircles, add these two points to the set of points E.

[0154] 4) Select two more points B and C from S and repeat step 3) until all point pairs are traversed. The point set E is the contour boundary.

[0155] Contour regularization: Directly connecting the boundary points of the contour extraction will result in a contour that is too rough, so the contour needs to be regularized. The Douglas-Peuker algorithm (DP algorithm) was proposed by David H. Douglas and Thomas K. Peucker in 1973 and is mainly used to simplify curves or polylines. The specific implementation steps are as follows

[0156] 1) Connect the first and last points A and B of the broken line to generate a straight line L.

[0157] 2) Calculate the distances from all points on the polyline to the straight line L in sequence.

[0158] 3) Set a distance threshold t. If the maximum distance D among all distances is less than the distance threshold t, delete all points on the polyline except A and B; if D is greater than the threshold t, split the polyline into two segments from the maximum distance point.

[0159] 4) Repeat steps 2) and 3) for all polyline segments until all distances D are less than the threshold.

[0160] The result after contour regularization is as follows Figure 9 As shown, the traffic marking feature point data can be extracted.

[0161] The accuracy detection module 304 is used to perform position accuracy detection processing using the traffic marking feature point data based on the same accuracy detection method to obtain a data accuracy detection result.

[0162] In the specific implementation process of the present invention, the same-precision detection method uses the traffic marking feature point data to perform position accuracy detection processing to obtain data accuracy detection results, including: selecting a preset number of corner points as checkpoints at intervals along the road direction, and obtaining the real coordinate data of the checkpoints; obtaining the point cloud coordinate data corresponding to the checkpoint based on the traffic marking feature point data, and using the real coordinate data and point cloud coordinate data of the checkpoint based on the same-precision detection method to perform position accuracy detection processing to obtain data accuracy detection results.

[0163] Specifically, the same precision detection method is used to perform precision detection on the vehicle-mounted laser point cloud data. The corner points of the road markings are obvious feature points. One or two corner points are selected at a certain distance along the road direction as checkpoints. The true coordinate values ​​of the checkpoints can be collected using the field RTK method. The point cloud coordinate values ​​corresponding to the checkpoints can be obtained by extracting the feature points of the road traffic markings. With the checkpoint as the center, the road traffic marking feature points within a certain radius are searched, and the difference between the two is calculated to detect the point accuracy; the calculation formula is as follows:

[0164] ;

[0165] ;

[0166] in, represents the error in the plane of the checkpoint; Indicates the error in the elevation of the checkpoint; Indicates the total number of checkpoints; Represents the difference between the true value of check point i and the point cloud feature point value in the X direction; Indicates the difference between the true value of the check point i and the point cloud feature point value in the Y direction; Indicates the elevation value of the point cloud feature point; Indicates the true value of the elevation of the checkpoint.

[0167] Experiments and analyses were conducted on the above-mentioned precision detection. To verify the feasibility and efficiency of this precision detection method, vehicle-mounted laser point cloud data containing different road scenes within an area of ​​10 square kilometers were selected for testing. The Tianbao vehicle-mounted 3D laser scanner MX50 was used to obtain point cloud data of the main roads and secondary roads in the test area. A total of 44 mobile measurements were completed in the survey area, with a total route of approximately 151.13 kilometers and an effective route of approximately 129.58 kilometers.

[0168] In order to detect the position accuracy of the vehicle-mounted point cloud, a total of 148 checkpoints were selected, and the points were all selected as road traffic marking feature points. The method proposed in this paper was used for detection, and the point cloud feature points corresponding to all checkpoints could be extracted. By comparing the checkpoints and point cloud feature points, the point cloud accuracy could be accurately detected. The accuracy calculation results are shown in Table 2.

[0169] Table 2 Point cloud accuracy statistics

[0170]

[0171] Note: The first 4 digits of both X and Y coordinates are removed.

[0172] Compared with traditional accuracy detection methods, the vehicle-mounted laser point cloud data position accuracy detection method based on road traffic marking feature points has obvious efficiency advantages; if the traditional method is used, it is necessary to manually select the feature points corresponding to the checkpoints in the point cloud data, which is time-consuming and labor-intensive, and has a low degree of automation; one engineer needs to spend 48 hours to complete the accuracy detection of 129.58 kilometers of vehicle-mounted point clouds in the test area; if the new method is used, using a graphics workstation (Core i9 processor, 32G memory, NVIDIA Quadro RTX4000 graphics card) to process point cloud data, one engineer only needs 27 hours to complete the accuracy detection of vehicle-mounted point clouds in the test area, and all processing is done automatically by computer; if a more powerful computing platform is used, it is possible to jointly process point cloud data of multiple roads, further improving detection efficiency.

[0173] In an embodiment of the present invention, data collection and processing are performed based on a vehicle-mounted laser device to obtain vehicle-mounted laser point cloud data; noise reduction pre-processing is performed on the vehicle-mounted laser point cloud data to obtain road surface point cloud data; marking point cloud segmentation processing is performed on the road surface point cloud data based on a preset algorithm, and the segmented marking point cloud data is subjected to traffic marking feature point extraction processing to obtain traffic marking feature point data; position accuracy detection processing is performed using the traffic marking feature point data based on the same accuracy detection method to obtain data accuracy detection results; automated accuracy detection is achieved, which greatly reduces the workload during detection, improves the accuracy of detection, and realizes a wider range of applications.

[0174] An embodiment of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the vehicle-mounted laser point cloud data position accuracy detection method described in any of the above-described embodiments. The computer-readable storage medium includes, but is not limited to, any type of disk (including floppy disks, hard disks, optical disks, CD-ROMs, and magneto-optical disks), ROM (Read-Only Memory), RAM (Random Access Memory), EPROM (Erasable Programmable Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory, magnetic cards, or optical cards. In other words, a storage device includes any medium that can store or transmit information in a readable form by a device (e.g., a computer or mobile phone), and can be a read-only memory, a magnetic disk, or an optical disk.

[0175] An embodiment of the present invention further provides a computer application program that runs on a computer and is used to execute the vehicle-mounted laser point cloud data position accuracy detection method of any one of the above embodiments.

[0176] also, Figure 4 It is a schematic diagram of the structure of an electronic device in an embodiment of the present invention.

[0177] The embodiment of the present invention further provides an electronic device, such as Figure 4 The electronic device includes a processor 402, a memory 403, an input unit 404, a display unit 405 and other components. Those skilled in the art will understand that Figure 4The structural components of the electronic device shown do not constitute a limitation on all devices, and may include more or fewer components than shown, or combine certain components. The memory 403 can be used to store the application 401 and various functional modules, and the processor 402 runs the application 401 stored in the memory 403, thereby executing various functional applications and data processing of the device. The memory can be an internal memory or an external memory, or include both internal and external memories. The internal memory may include a read-only memory (ROM), a programmable ROM (PROM), an electrically programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), a flash memory, or a random access memory. The external memory may include a hard disk, a floppy disk, a ZIP disk, a USB flash drive, a magnetic tape, etc. The memory disclosed in the present invention includes but is not limited to these types of memories. The memory disclosed in the present invention is only an example and not a limitation.

[0178] The input unit 404 is used to receive input signals and keywords entered by the user. The input unit 404 may include a touch panel and other input devices. The touch panel can detect user touch operations on or near it (e.g., operations performed on or near the touch panel using a finger, stylus, or any other suitable object or accessory) and activate corresponding connected devices according to pre-set programs. Other input devices may include, but are not limited to, one or more of a physical keyboard, function keys (e.g., playback control keys, on / off buttons, etc.), a trackball, a mouse, and a joystick. The display unit 405 is used to display user input or information provided to the user, as well as various menus of the terminal device. The display unit 405 may be in the form of a liquid crystal display (LCD), an organic light-emitting diode (OLED), or other devices. The processor 402 is the control center of the terminal device. It connects various components of the device using various interfaces and circuits. It performs various functions and processes data by running or executing software programs and / or modules stored in the memory 403 and accessing data stored in the memory.

[0179] As an embodiment, the electronic device includes: one or more processors 402, a memory 403, and one or more applications 401, wherein the one or more applications 401 are stored in the memory 403 and are configured to be executed by the one or more processors 402, and the one or more applications 401 are configured to execute the corresponding vehicle-mounted laser point cloud data position accuracy detection method in any of the above embodiments.

[0180] In an embodiment of the present invention, data collection and processing are performed based on a vehicle-mounted laser device to obtain vehicle-mounted laser point cloud data; noise reduction pre-processing is performed on the vehicle-mounted laser point cloud data to obtain road surface point cloud data; marking point cloud segmentation processing is performed on the road surface point cloud data based on a preset algorithm, and the segmented marking point cloud data is subjected to traffic marking feature point extraction processing to obtain traffic marking feature point data; position accuracy detection processing is performed using the traffic marking feature point data based on the same accuracy detection method to obtain data accuracy detection results; automated accuracy detection is achieved, which greatly reduces the workload during detection, improves the accuracy of detection, and realizes a wider range of applications.

[0181] In addition, the above is a detailed introduction to a vehicle-mounted laser point cloud data position accuracy detection method based on road traffic marking feature points and related devices provided by an embodiment of the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for detecting the position accuracy of vehicle-mounted laser point cloud data based on road traffic marking feature points, characterized in that: The method comprises: Carry out data collection and processing based on vehicle-mounted laser equipment to obtain vehicle-mounted laser point cloud data; Performing noise reduction preprocessing on the vehicle-mounted laser point cloud data to obtain road surface point cloud data; Performing traffic marking point cloud segmentation processing on the road surface point cloud data based on a preset algorithm, and performing traffic marking feature point extraction processing on the segmented traffic marking point cloud data to obtain traffic marking feature point data; Based on the same precision detection method, the traffic marking feature point data is used to perform position accuracy detection processing to obtain a data accuracy detection result; The segmented traffic marking point cloud data is subjected to traffic marking feature point extraction processing to obtain traffic marking feature point data, including: Set each segmentation mark point cloud data as set S, set the boundary point set E, and set the boundary point set E to be an empty set, and set the radius threshold to A; Thus, if we randomly select two points B and C from the set S, and the distance between B and C is not greater than 2A, we generate two circumscribed circles with radius A that pass through both B and C. If there are no other points in set S within any of the two circumscribed circles, then points B and C are added to the boundary point set E. This continues until all point pairs in set S are traversed, and the point cloud data in the boundary point set E is used as the outline boundary of the traffic marking. Performing contour regularization processing on the contour boundary, and obtaining traffic marking feature point data according to the contour regularization processing result; The method of performing position accuracy detection using the traffic marking feature point data based on the same accuracy detection method to obtain a data accuracy detection result includes: Selecting a preset number of corner points at intervals along the road direction as checkpoints, and obtaining real coordinate data of the checkpoints; The point cloud coordinate data corresponding to the checkpoint is obtained based on the traffic marking feature point data, and the position accuracy detection processing is performed using the real coordinate data and the point cloud coordinate data of the checkpoint based on the same accuracy detection method to obtain the data accuracy detection result.

2. The vehicle-mounted laser point cloud data position accuracy detection method according to claim 1, characterized in that: The performing noise reduction preprocessing on the vehicle-mounted laser point cloud data to obtain road surface point cloud data includes: performing a denoising process on the vehicle-mounted laser point cloud data by filtering out non-ground point cloud data based on a cloth simulation filtering algorithm to obtain vehicle-mounted laser point cloud data after a first denoising, wherein the cloth simulation filtering algorithm is an algorithm set using preset cloth rigidity parameters and preset cloth grid resolution parameters; The road surface point cloud data is extracted and processed based on the scanning line algorithm from the vehicle-mounted laser point cloud data after the first noise reduction to obtain the road surface point cloud data.

3. The vehicle-mounted laser point cloud data position accuracy detection method according to claim 2, characterized in that: The scanning line algorithm is used to extract and process the road surface point cloud data from the vehicle-mounted laser point cloud data after the first noise reduction to obtain the road surface point cloud data, including: Based on the scanning line algorithm, the horizontal distance and slope difference of the point cloud within a fixed window on each laser scanning line are used to find the road bump points. All the road bump points in the vehicle-mounted laser point cloud data after the first noise reduction are merged into the road bump points on the left and right sides. The road bump points on the same side are sequentially connected to form a road edge line, and the road edge line is used to form a road contour. The road contour is then used to crop and extract the ground point cloud in the vehicle-mounted laser point cloud data after the first noise reduction to obtain road surface point cloud data.

4. The method for detecting the position accuracy of vehicle-mounted laser point cloud data according to claim 1, characterized in that: The performing line marking point cloud segmentation processing on the road surface point cloud data based on a preset algorithm includes: Performing clustering processing on the road surface point cloud data based on a K-means clustering algorithm to obtain a first clustering result; Dividing the road surface point cloud data into road traffic marking point cloud data and non-road traffic marking point cloud data based on the first clustering result; The road traffic marking point cloud data is subjected to denoising and optimization processing, and the denoised and optimized road traffic marking point cloud data is segmented into an independent road marking point cloud object based on a Euclidean distance algorithm to form segmented road marking point cloud data.

5. The vehicle-mounted laser point cloud data position accuracy detection method according to claim 4, characterized in that: The Euclidean distance algorithm segments the denoised and optimized road traffic marking point cloud data into an independent road marking point cloud object to form segmented road marking point cloud data, including: Initialize the point set Q so that the initialized point set Q is an empty set; Randomly select any point in the denoised and optimized road traffic marking point cloud data as the initial point P; Perform a proximity search on the initial point P to obtain all points within the set distance threshold r of the initial point P and add them to Q; Perform a neighbor search on all points in the point set Q that have not been searched for according to the set distance threshold r, and obtain a cluster of line point cloud data within the set distance threshold r; Until all points of the denoised and optimized road traffic marking point cloud data are traversed, segmented marking point cloud data is formed.

6. A vehicle-mounted laser point cloud data position accuracy detection device based on road traffic marking feature points, characterized in that: The device comprises: Data acquisition module: used to collect and process data based on vehicle-mounted laser equipment to obtain vehicle-mounted laser point cloud data; Data denoising module: used for performing denoising preprocessing on the vehicle-mounted laser point cloud data to obtain road surface point cloud data; Feature point extraction module: used to perform traffic marking point cloud segmentation processing on the road surface point cloud data based on a preset algorithm, and segment the traffic marking point cloud data to perform traffic marking feature point extraction processing to obtain traffic marking feature point data; Accuracy detection module: used to perform position accuracy detection processing using the traffic marking feature point data based on the same accuracy detection method to obtain data accuracy detection results; The segmented traffic marking point cloud data is subjected to traffic marking feature point extraction processing to obtain traffic marking feature point data, including: Set each segmentation mark point cloud data as set S, set the boundary point set E, and set the boundary point set E to be an empty set, and set the radius threshold to A; Thus, if we randomly select two points B and C from the set S, and the distance between B and C is not greater than 2A, we generate two circumscribed circles with radius A that pass through both B and C. If there are no other points in set S within any of the two circumscribed circles, then points B and C are added to the boundary point set E. This continues until all point pairs in set S are traversed, and the point cloud data in the boundary point set E is used as the outline boundary of the traffic marking. Performing contour regularization processing on the contour boundary, and obtaining traffic marking feature point data according to the contour regularization processing result; The method of performing position accuracy detection using the traffic marking feature point data based on the same accuracy detection method to obtain a data accuracy detection result includes: Selecting a preset number of corner points at intervals along the road direction as checkpoints, and obtaining real coordinate data of the checkpoints; The point cloud coordinate data corresponding to the checkpoint is obtained based on the traffic marking feature point data, and the position accuracy detection processing is performed using the real coordinate data and the point cloud coordinate data of the checkpoint based on the same accuracy detection method to obtain the data accuracy detection result.

7. An electronic device comprising a processor and a memory, characterized in that: The processor runs the computer program or code stored in the memory to implement the vehicle-mounted laser point cloud data position accuracy detection method according to any one of claims 1 to 5.

8. A computer-readable storage medium for storing a computer program or code, characterized in that: When the computer program or code is executed by a processor, the method for detecting the position accuracy of vehicle-mounted laser point cloud data according to any one of claims 1 to 5 is implemented.

Images