Lane line point cloud data supplementing method and device, electronic equipment and vehicle

Abstract

The invention relates to the technical field of automatic driving, in particular to a lane line point cloud data supplementing method and device, electronic equipment and a vehicle. The method comprises the following steps: fitting obtained lane line point cloud data to obtain a fitted curve; determining a point cloud sparse region of the lane line point cloud data; determining a target curve segment coinciding with the point cloud sparse region on the fitting curve; encrypting and sampling the target curve segment to obtain an encrypted sampling point in the target curve segment; performing dimension transformation on the encrypted sampling points through lane line point cloud data to obtain three-dimensional encrypted sampling points; and placing the three-dimensional encrypted sampling point into the lane line point cloud data to obtain target lane line point cloud data. Through the method, fitting supplement of the lane line point cloud data can be carried out by using crowdsourcing point cloud data or other point cloud data, the lane line point cloud data can be processed more finely to obtain more accurate target lane line point cloud data, and then an automatic driving high-precision map is produced.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous driving, and particularly to a method, device, electronic device and vehicle for supplementing lane line point cloud data. Background Art

[0002] In order to meet the high-precision map requirements of autonomous driving, three-dimensional point cloud road data fusion is usually performed by globally transforming and registering data collected multiple times. However, for lane line data, due to the complexity of the actual road conditions and lane line models, the accuracy of the data after global transformation may be insufficient. A common method for lane line extraction is to create a binary image of gray scale by using the average gray scale of the points within a grid, and then extract the lane line model through clustering. However, this method performs poorly on data collected in the crowdsourcing mode because the quality and accuracy of this data are usually low. When the points within the grid are sparsely distributed, the established binary image of lane line gray scale may be unstable. Therefore, a more refined method is needed to process lane line point cloud data to extract accurate lane lines. Summary of the Invention

[0003] In view of the above problems, embodiments of the present invention provide a method, device, electronic device and vehicle for supplementing lane line point cloud data, so as to overcome or at least partially solve the above problems.

[0004] In a first aspect, an embodiment of the present invention provides a method for supplementing lane line point cloud data, the method including:

[0005] Obtaining a fitting curve of the lane line point cloud data by fitting the acquired lane line point cloud data;

[0006] Determining a point cloud sparse region of the lane line point cloud data according to the lane line point cloud data;

[0007] Determining a target curve segment on the fitting curve that coincides with the point cloud sparse region according to the fitting curve and the point cloud sparse region;

[0008] Obtaining encrypted sampling points in the target curve segment by performing encrypted sampling on the target curve segment;

[0009] Performing dimensional transformation on the encrypted sampling points through the lane line point cloud data to obtain three-dimensional encrypted sampling points;

[0010] Placing the three-dimensional encrypted sampling points into the lane line point cloud data to obtain target lane line point cloud data.

[0011] Optionally, the determining the point cloud sparse region of the lane line point cloud data according to the lane line point cloud data includes:

[0012] Determine the circumscribed region of the lane line point cloud data according to the lane line point cloud data;

[0013] Divide the circumscribed region into multiple sub-regions according to a set step size;

[0014] Determine the point cloud sparse regions in the multiple sub-regions according to a point cloud density threshold.

[0015] Optionally, the determining the point cloud sparse regions in the multiple sub-regions according to a point cloud density threshold includes:

[0016] Determine the number of point clouds in each of the multiple sub-regions according to the lane line point cloud data;

[0017] Determine the point cloud sparse regions in the multiple sub-regions according to the number of point clouds in each of the multiple sub-regions and the point cloud density threshold.

[0018] Optionally, the determination of the point cloud density threshold includes:

[0019] Determine the number of point clouds in the circumscribed region according to the lane line point cloud data;

[0020] Determine the point cloud density threshold according to the circumscribed region and the number of point clouds.

[0021] Optionally, the obtaining the encrypted sampling points in the target curve segment by performing encrypted sampling on the target curve segment includes:

[0022] Determine the point cloud density of the point cloud sparse region according to the point cloud sparse region;

[0023] Determine the number of sampling times for performing encrypted sampling on the target curve segment according to the point cloud density and the point cloud density threshold;

[0024] Perform encrypted sampling on the target curve segment for the number of sampling times according to the number of sampling times to obtain the encrypted sampling points in the target curve segment.

[0025] Optionally, the obtaining three-dimensional encrypted sampling points by performing dimensional transformation on the encrypted sampling points through the lane line point cloud data includes:

[0026] Obtain the fitting surface of the lane line point cloud data by fitting the acquired lane line point cloud data;

[0027] Substitute the coordinates of the encrypted sampling points into the surface equation of the fitting surface to obtain three-dimensional encrypted sampling points.

[0028] Optionally, the method further includes:

[0029] Obtain multiple groups of target lane line point cloud data of the same lane line, and determine it as lane line cluster point cloud data;

[0030] By vectorizing the lane line cluster point cloud data, lane line vectorized data is obtained.

[0031] In a second aspect, an embodiment of the present invention proposes a lane line point cloud data supplement device, and the device includes:

[0032] A first obtaining module, configured to obtain a fitting curve of the lane line point cloud data by fitting the acquired lane line point cloud data;

[0033] A second determining module, configured to determine a point cloud sparse area of the lane line point cloud data according to the lane line point cloud data;

[0034] A third determining module, configured to determine a target curve segment on the fitting curve that coincides with the point cloud sparse area according to the fitting curve and the point cloud sparse area;

[0035] A fourth obtaining module, configured to obtain encrypted sampling points in the target curve segment by performing encrypted sampling on the target curve segment;

[0036] A fifth obtaining module, configured to perform dimensional transformation on the encrypted sampling points through the lane line point cloud data to obtain three-dimensional encrypted sampling points;

[0037] A sixth obtaining module, configured to place the three-dimensional encrypted sampling points into the lane line point cloud data to obtain target lane line point cloud data.

[0038] Optionally, the second determining module includes:

[0039] A first determining sub-module, configured to determine an external region of the lane line point cloud data according to the lane line point cloud data;

[0040] A second dividing sub-module, configured to divide the external region into multiple sub-regions according to a set step size;

[0041] A third determining sub-module, configured to determine a point cloud sparse area in the multiple sub-regions according to a point cloud density threshold.

[0042] Optionally, the third determining sub-module includes:

[0043] A fourth determining sub-module, configured to determine the number of point clouds in each of the multiple sub-regions according to the lane line point cloud data;

[0044] A fifth determining sub-module, configured to determine a point cloud sparse area in the multiple sub-regions according to the number of point clouds in each of the multiple sub-regions and the point cloud density threshold.

[0045] Optionally, the device further includes:

[0046] A seventh determination module, configured to determine the number of point clouds in the circumscribed area according to the lane line point cloud data;

[0047] An eighth determination module, configured to determine the point cloud density threshold according to the circumscribed area and the number of point clouds.

[0048] Optionally, the fourth acquisition module includes:

[0049] A sixth determination sub-module, configured to determine the point cloud density of the point cloud sparse area according to the point cloud sparse area;

[0050] A seventh determination sub-module, configured to determine the number of sampling times for encrypting and sampling the target curve segment according to the point cloud density and the point cloud density threshold;

[0051] An eighth acquisition sub-module, configured to perform encrypted sampling of the target curve segment for the number of sampling times according to the number of sampling times, and obtain encrypted sampling points in the target curve segment.

[0052] Optionally, the fifth acquisition module includes:

[0053] A ninth acquisition sub-module, configured to obtain a fitting surface of the lane line point cloud data by fitting the acquired lane line point cloud data;

[0054] A tenth acquisition sub-module, configured to substitute the coordinates of the encrypted sampling points into the surface equation of the fitting surface to obtain three-dimensional encrypted sampling points.

[0055] Optionally, the device further includes:

[0056] A ninth determination module, configured to acquire multiple groups of target lane line point cloud data of the same lane line and determine them as lane line cluster point cloud data;

[0057] A tenth acquisition module, configured to obtain lane line vectorized data by vectorizing the lane line cluster point cloud data.

[0058] In a third aspect, an embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the electronic device executes the computer program to implement the lane line point cloud data supplement method according to any one of the first aspects of the embodiments of the present invention.

[0059] Fourthly, an embodiment of the present invention provides a vehicle, which includes an autonomous driving module for obtaining lane line point cloud data determined by the lane line point cloud data supplement method according to any one of the first aspects of the embodiments of the present invention.

[0060] Advantages of the present invention:

[0061] The lane line point cloud data supplement method provided by the present invention fits the obtained lane line point cloud data to obtain a fitting curve of the lane line point cloud data; determines a point cloud sparse area of the lane line point cloud data according to the lane line point cloud data; determines a target curve segment on the fitting curve that coincides with the point cloud sparse area according to the fitting curve and the point cloud sparse area; obtains encrypted sampling points in the target curve segment by performing encrypted sampling on the target curve segment; performs dimensional transformation on the encrypted sampling points through the lane line point cloud data to obtain three-dimensional encrypted sampling points; and places the three-dimensional encrypted sampling points into the lane line point cloud data to obtain target lane line point cloud data. Through the lane line point cloud data supplement method provided by the present invention, crowd-sourced point cloud data or other point cloud data can be used to perform fitting and supplement of the lane line point cloud data, and the lane line point cloud data can be processed more precisely to obtain more accurate target lane line point cloud data, and then an autonomous driving high-precision map can be produced. Description of the Drawings

[0062] Figure 1 is a flowchart of the steps of a lane line point cloud data supplement method proposed in an embodiment of the present invention;

[0063] Figure 2 is a schematic diagram of the fitting curves of two sets of lane line point cloud data included in the same lane line proposed in an embodiment of the present invention;

[0064] Figure 3 is a schematic diagram of the sampling of the fitting curve of the lane line point cloud data and the encrypted sampling of the target curve segment proposed in an embodiment of the present invention;

[0065] Figure 4 is a schematic diagram of the circumscribed rectangle and the detection sliding window in the two-dimensional plane of the lane line point cloud data proposed in an embodiment of the present invention;

[0066] Figure 5 is a schematic diagram of the fitting surface of the lane line point cloud data proposed in an embodiment of the present invention;

[0067] Figure 6 is a schematic diagram of the main steps of the lane line point cloud data supplement method proposed in an embodiment of the present invention;

[0068] Figure 7Function module diagram of a lane line point cloud data supplement system proposed in an embodiment of the present invention;

[0069] Figure 8 Block diagram of a module of a lane line point cloud data supplement device proposed in an embodiment of the present invention. Specific implementation manner

[0070] The following will illustrate the implementation manners of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand the other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific implementation manners. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, rather than for limiting the protection scope of the present invention.

[0071] It should be noted that the diagrams provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner. Therefore, only the components related to the present invention are shown in the diagrams, rather than being drawn according to the number, shape, and size of the components in actual implementation. The types, quantities, and proportions of the components in actual implementation can be arbitrarily changed, and the component layout type may also be more complex.

[0072] In a first aspect, an embodiment of the present invention proposes a method for supplementing lane line point cloud data, as Figure 1 shown, Figure 1 is a step flow chart of a method for supplementing lane line point cloud data proposed in an embodiment of the present invention. The method includes the following steps:

[0073] Step S101: By fitting the acquired lane line point cloud data, obtain the fitting curve of the lane line point cloud data.

[0074] In this embodiment, the lane line point cloud data can be obtained from the crowdsourced point cloud data. Specifically, first, the crowdsourced point cloud data is obtained, and then through semantic segmentation, the initial lane line point cloud data is obtained. The initial lane line point cloud data contains the initial point cloud data of multiple lane lines, and for one lane line, there may also be multiple groups of initial lane line point cloud data. For example, the initial lane line point cloud data may include the initial point cloud data of lane lines C1, C2, C3... Cn, and the initial point cloud data of lane line C1 may also include group R1 initial lane line point cloud data, group R2 initial lane line point cloud data, group R3 initial lane line point cloud data... group Rn initial lane line point cloud data. The order of each group of lane line point cloud data in the initial lane line point cloud data is chaotic. Therefore, it is necessary to perform denoising processing on the obtained initial lane line point cloud data, and the denoising processing can adopt the method of statistical filtering denoising. Then, perform spatial clustering on the denoised initial lane line point cloud data, classify the initial lane line point cloud data according to the lane lines, and then the one or more groups of lane line point cloud data corresponding to a single lane line can be determined. Optionally, the algorithm used for spatial clustering can be the DBSCAN algorithm for clustering. After obtaining the lane line point cloud data in this way, it is possible to fit each group of lane line point cloud data obtained, that is, using the least squares polynomial method, project each group of lane line point cloud data onto a two-dimensional plane, and fit the curve of each group of lane line point cloud data. Specifically, in this example, a quadratic polynomial of one variable is used to obtain the fitting curve y = a1x 2 + a2x + a3, where: x and y are the abscissa and ordinate of each point on the fitting curve respectively, and a1, a2, and a3 are constants. As Figure 2 shown Figure 2 is a schematic diagram of the fitting curves of two groups of lane line point cloud data included in the same lane line proposed in an embodiment of the present invention.

[0075] Step S102: Determine the point cloud sparse region of the lane line point cloud data according to the lane line point cloud data.

[0076] In this embodiment, the point cloud sparse region is the region where the number of point clouds (or point cloud density) in the lane line point cloud data is less than the threshold (or average value). According to the lane line point cloud data, information such as the number of point clouds and point cloud density in the lane line point cloud data can be obtained; the lane line point cloud data can also be divided into several regions to determine the point cloud sparse region. In this embodiment, the specific method for determining the point cloud sparse region of the lane line point cloud data is not limited. Optionally, a point cloud number threshold or a point cloud density threshold can be set by oneself, or it can be flexibly selected according to the number of point clouds or point cloud density of a single group of lane line point cloud data.

[0077] Step S103: Determine a target curve segment on the fitting curve that coincides with the sparse point cloud region according to the fitting curve and the sparse point cloud region.

[0078] In this embodiment, after determining the sparse point cloud region of the lane line point cloud data, it is necessary to find the corresponding target curve segment on the fitting curve of the lane line point cloud data, that is, determine the target curve segment that coincides with the sparse point cloud region on the fitting curve. For example, if the sparse point cloud regions of the lane line point cloud data are region B and region F, then it is necessary to determine the target curve segment B' and the target curve segment F' that coincide with the sparse point cloud regions on the fitting curve.

[0079] Step S104: Obtain encrypted sampling points in the target curve segment by performing encrypted sampling on the target curve segment.

[0080] In this embodiment, encrypted sampling is performed on the target curve segment that coincides with the sparse point cloud region on the fitting curve. Sampling is to obtain the horizontal and vertical coordinates of a certain point (sampling point) on the target curve segment. Encrypted sampling is to obtain more sampling points to make the sampling points denser. Specifically, uniform sampling with a fixed step size can be first performed on the entire fitting curve, and then the sampling points in the target curve segment on the fitting curve are encrypted, that is, sampling points are added in the target curve segment to make the sampling points denser. The final obtained all sampling points are encrypted sampling points, as Figure 3 shown. Figure 3 FIG. is a schematic diagram of sampling of a fitting curve of lane line point cloud data and encrypted sampling of a target curve segment according to an embodiment of the present invention. Specifically, the method of encrypted sampling can be to sample the midpoints of adjacent sampling points obtained by uniform sampling in the target curve segment, and then sample the midpoints of the new adjacent sampling points; it can also perform uniform sampling in the target curve segment according to a fixed step size. The specific encrypted sampling method is not limited in this embodiment.

[0081] Optionally, encrypted sampling is not infinite sampling. The number of encrypted sampling times can be set to obtain a fixed number of encrypted sampling points. For example, 3 times of encrypted sampling can be performed on each target curve segment. The number of encrypted sampling times can also be determined according to the sparsity of the corresponding sparse point cloud region. For example, the target curve segment A originally contains 3 point clouds, and the target curve segment B originally contains 5 point clouds. Then 4 times of encrypted sampling can be performed on the target curve segment A and 2 times of encrypted sampling can be performed on the target curve segment B. Of course, it can also be determined according to other methods for determining the number of encrypted sampling times, such as according to the point cloud density of each target curve segment, etc.

[0082] Step S105: Perform dimensional transformation on the encrypted sampling points through the lane line point cloud data to obtain three-dimensional encrypted sampling points.

[0083] In this embodiment, the encrypted sampling points are obtained from the target curve segment on the fitting curve, and the fitting curve is a two-dimensional curve. Therefore, the encrypted sampling points only contain horizontal and vertical coordinates. Therefore, it is necessary to perform dimensional transformation on the encrypted sampling points through the lane line point cloud data to obtain the three-dimensional coordinates of the encrypted sampling points, that is, to obtain three-dimensional encrypted sampling points.

[0084] Step S106: Place the three-dimensional encrypted sampling points into the lane line point cloud data to obtain the target lane line point cloud data.

[0085] In this embodiment, after obtaining the three-dimensional coordinates of the encrypted sampling points, that is, obtaining the three-dimensional encrypted sampling points, the three-dimensional encrypted sampling points can be supplemented into the lane line point cloud data to make the point cloud in the lane line point cloud data denser, obtain the target lane line point cloud data, and achieve the purpose of supplementing the lane line point cloud data.

[0086] Specifically, through the lane line point cloud data supplement method provided by the present invention, crowd-sourced point cloud data or other point cloud data can be used to perform fitting and supplement of the lane line point cloud data, and the lane line point cloud data can be processed more precisely to obtain more accurate target lane line point cloud data, and then produce an autonomous driving high-precision map.

[0087] Optionally, the step S102 includes:

[0088] Step S1021: Determine the circumscribed region of the lane line point cloud data according to the lane line point cloud data.

[0089] In this embodiment, according to the lane line point cloud data, the maximum coordinates and minimum coordinates of all lane line point cloud data in the lane line point cloud data can be determined, and then the circumscribed region of the lane line point cloud data can be determined. Specifically, Figure 4 is a schematic diagram of the circumscribed rectangle and detection sliding window of the two-dimensional plane of the lane line point cloud data proposed in an embodiment of the present invention. As Figure 4 shown, after obtaining the fitting curve of the lane line point cloud data, according to the lane line point cloud data, the maximum and minimum values of the horizontal and vertical coordinates on the fitting curve can be determined, that is, x min 、x max 、y min 、y max . Then, according to the maximum and minimum values of the horizontal and vertical coordinates on the fitting curve, the circumscribed rectangle of the fitting curve is determined, that is, the circumscribed region of the lane line point cloud data in the two-dimensional plane is determined. Optionally, the maximum and minimum values of the horizontal, vertical, and vertical coordinates of the lane line point cloud data can also be directly determined according to the three-dimensional coordinates of the lane line point cloud data, and then the circumscribed region of the lane line point cloud data can be determined.

[0090] Step S1022: Divide the circumscribed region into multiple sub-regions according to a set step size.

[0091] In this embodiment, it is necessary to divide the external region into multiple sub-regions according to a set step size, where the multiple sub-regions can overlap. Specifically, as Figure 4 shown, the external region can be divided into multiple sub-regions by using a sliding window. For example, the window length is set to 1 meter, the window forward step size is 0.5 meter, and the window width is the width of the external rectangle (y max -y min ), and then the external region can be divided into multiple sub-regions with a length of 1 meter and a width of the width of the external rectangle (y max -y min ), and adjacent sub-regions overlap each other. For example, the coordinates of the endpoints of three adjacent sub-regions are shown in Table 1:

[0092] Table 1

[0093] Endpoint 1 Endpoint 2 Endpoint 3 Endpoint 4 Sub-region A (2,4) (3,4) (2,1) (3,1) Sub-region B (2.5,4) (3.5,4) (2.5,1) (3.5,1) Sub-region C (3,4) (4,4) (3,1) (4,1)

[0094] Step S1023: Determine the point cloud sparse regions in the multiple sub-regions according to the point cloud density threshold.

[0095] In this embodiment, the point cloud density threshold can be preset according to test data or determined according to the current lane line point cloud data, which is not limited in this embodiment. By comparing the point cloud quantity or point cloud density of each sub-region with the point cloud density threshold, it can be determined which of the sub-regions in each sub-region belong to the point cloud sparse regions.

[0096] Specifically, through the method provided in this embodiment, the point cloud sparse regions in the lane line point cloud data can be accurately determined to meet the subsequent accurate supplementation of the lane line point cloud data.

[0097] Optionally, step S1023 includes:

[0098] Step S10231: Determine the point cloud quantity of each of the multiple sub-regions according to the lane line point cloud data.

[0099] In this embodiment, according to the point cloud quantity in the lane line point cloud data, the point cloud quantity of each of the multiple sub-regions can be determined.

[0100] Step S10232: Determine the point cloud sparse regions in the multiple sub-regions according to the point cloud quantity of each of the multiple sub-regions and the point cloud density threshold.

[0101] In this embodiment, according to the number of point clouds in each of the multiple sub-regions, the point cloud density of each of the multiple sub-regions can be determined. Specifically, in combination with sub-region A, sub-region B, and sub-region C in the previous example, assuming that there are 5 point cloud data in sub-region A, 7 point cloud data in sub-region B, and 4 point cloud data in sub-region C, and the point cloud density threshold is 2 per square meter, the determination results are shown in Table 2:

[0102] Table 2

[0103] Endpoint 1 Endpoint 2 Endpoint 3 Endpoint 4 Number of point clouds Point cloud density Sparse area of point cloud Sub-region A (2,4) (3,4) (2,1) (3,1) 5 5 / 3＝1.67＜2 Yes Sub-region B (2.5,4) (3.5,4) (2.5,1) (3.5,1) 7 7 / 3＝2.33＞2 No Sub-region C (3,4) (4,4) (3,1) (4,1) 4 4 / 3＝1.33＜2 Yes

[0104] As shown in Table 2, sub-region A and sub-region C are sparse point cloud regions, and sub-region B is not a sparse point cloud region.

[0105] Optionally, the determination of the point cloud density threshold includes the following steps:

[0106] Step S107: Determine the number of point clouds in the circumscribed region according to the lane line point cloud data.

[0107] Step S108: Determine the point cloud density threshold according to the circumscribed region and the number of point clouds.

[0108] In this embodiment, first, according to the lane line point cloud data, the number of point clouds in the circumscribed region, that is, the number of point clouds in the lane line point cloud data, is determined. Then, according to the area (or volume) of the circumscribed region and the number of point clouds, the point cloud density threshold is determined. Specifically, if the circumscribed region is a two-dimensional circumscribed region, the point cloud density threshold is determined according to the area of the circumscribed region and the number of point clouds; if the circumscribed region is a three-dimensional circumscribed region, the point cloud density threshold is determined according to the volume of the circumscribed region and the number of point clouds. For example, after fitting each group of obtained lane line point cloud data, that is, using the least square polynomial method to project the lane line point cloud data onto a two-dimensional plane and fitting the fitting curve of the lane line point cloud data, according to the area of the circumscribed rectangle of the fitting curve and the number of point clouds in the circumscribed rectangle, the average point cloud density in the circumscribed rectangle can be calculated, and then the obtained average point cloud density is determined as the point cloud density threshold.

[0109] Optionally, step S104 includes:

[0110] Step S1041: Determine the point cloud density of the point cloud sparse region according to the point cloud sparse region.

[0111] Step S1042: Determine the sampling times for encrypting and sampling the target curve segment according to the point cloud density and the point cloud density threshold.

[0112] In this embodiment, after determining the sparse point cloud region, it is first necessary to determine the point cloud density of the sparse point cloud region; then, based on the difference between the point cloud density of the sparse point cloud region and the point cloud density threshold, determine the number of sampling times for encrypting and sampling the target curve segment. Specifically, determining the number of sampling times for encrypting and sampling the target curve segment is to ensure that the point cloud density in the sparse point cloud region after encryption and sampling can be greater than or equal to the point cloud density threshold, that is, to make the sparse point cloud region no longer a sparse point cloud region after encryption and sampling. For example, assuming the point cloud density threshold is 2 per square meter, the encryption and sampling times are shown in Table 3:

[0113] Table 3

[0114] Number of point clouds Point cloud density Sparse area of point cloud Number of encrypted samplings Point cloud density after encryption Sub-region A 5 5 / 3＝1.67＜2 Yes 1 (5+1) / 3＝2 Sub-region B 7 7 / 3＝2.33＞2 No 0 7 / 3＝2.33 Sub-region C 4 4 / 3＝1.33＜2 Yes 2 (4+2) / 3＝2

[0115] As shown in Table 3, to ensure that the point cloud density after encryption is greater than or equal to the point cloud density threshold, sub-region A is encrypted and sampled 1 time, sub-region B is not encrypted and sampled, and sub-region C is encrypted and sampled 2 times. Optionally, it is also possible to make the point cloud density after encryption greater than the point cloud density threshold, then sub-region A needs to be encrypted and sampled 2 times, and sub-region C needs to be encrypted and sampled 3 times.

[0116] Step S1043: According to the number of sampling times, perform the encryption and sampling of the target curve segment for the number of sampling times to obtain the encrypted sampling points in the target curve segment.

[0117] In this embodiment, the method of encryption and sampling can be midpoint sampling, that is, sampling at the midpoint of the target curve segment; it can also be sampling according to a preset step size or in proportion. For example, if sub-region A needs to be sampled 1 time, the minimum abscissa of sub-region A is 2, and the maximum abscissa is 3, then sampling can be performed at the midpoint of sub-region A, that is, at the abscissa of 2.5. Substitute the abscissa = 2.5 into the curve equation of the fitting curve to obtain the ordinate of the sampling point, and then the encrypted sampling points in sub-region A can be obtained. Another example, if sub-region C needs to be sampled 2 times, the minimum abscissa of sub-region C is 3, and the maximum abscissa is 4, then sampling can be performed in proportion at the abscissas of 3.3 and 3.6. Substitute the abscissas = 3.3 and 3.5 into the curve equation of the fitting curve respectively to obtain the ordinates of the two sampling points, and then the encrypted sampling points in sub-region A can be obtained.

[0118] Optionally, step S105 includes:

[0119] Step S1051: By fitting the obtained lane line point cloud data, obtain the fitting surface of the lane line point cloud data.

[0120] In this embodiment, the method of obtaining the fitting surface of the lane line point cloud data can be the least squares method. As Figure 5 shown Figure 5Schematic diagram of a fitted surface of lane line point cloud data proposed in an embodiment of the present invention. The fitted surface Z(x, y) = ax of the lane line point cloud data is calculated by the least squares method 2 +bxy+cy 2 +dx+ey, where x, y, and z are the horizontal, vertical, and vertical coordinates of each point on the fitted surface respectively, and a, b, c, d, and e are constants.

[0121] Step S1052: Substitute the coordinates of the encrypted sampling points into the surface equation of the fitted surface to obtain three-dimensional encrypted sampling points.

[0122] In this embodiment, by substituting the coordinates (i.e., horizontal and vertical coordinates) of the encrypted sampling points into the surface equation of the fitted surface, the vertical coordinates of the encrypted sampling points can be obtained, and then three-dimensional encrypted sampling points can be obtained.

[0123] Optionally, the method further includes the following steps:

[0124] Step S109: Obtain multiple groups of target lane line point cloud data of the same lane line and determine them as lane line cluster point cloud data.

[0125] In this embodiment, after supplementing multiple groups of lane line point cloud data of the same lane line and obtaining each group of target lane line point cloud data, the multiple groups of target lane line point cloud data of the same lane line can be determined as lane line cluster point cloud data.

[0126] Step S110: Obtain lane line vectorized data by vectorizing the lane line cluster point cloud data.

[0127] In this embodiment, the lane line cluster point cloud data is vectorized to obtain lane line vectorized data for generating a high-precision map for autonomous driving.

[0128] Specifically, in a preferred implementation manner, as Figure 6 shown, Figure 6The figure is a schematic flow chart of the main steps of a method for supplementing lane line point cloud data proposed in an embodiment of the present invention. First, obtain the lane line point cloud data after semantic segmentation; then perform denoising processing on each group of lane line point cloud data; obtain the lane line point cloud data of each lane through spatial clustering; then project the lane line point cloud data onto a two-dimensional plane to obtain the fitting curve of the lane line point cloud data, and at the same time establish a circumscribed rectangle; use a sliding window to determine the sparse area of the point cloud; if there is no sparse area of the point cloud, perform uniform sampling on the fitting curve to obtain sampling points; if there is a sparse area of the point cloud, after performing uniform sampling on the fitting curve, perform encrypted sampling on the sparse area of the point cloud to obtain encrypted sampling points; then calculate the fitting surface of the lane line point cloud data, and calculate the three-dimensional encrypted sampling points through the surface equation of the fitting surface for the encrypted sampling points (sampling points); supplement the three-dimensional encrypted sampling points to the lane line point cloud data to obtain the target lane line point cloud data.

[0129] After obtaining the target lane line point cloud data, it is also necessary to perform quality evaluation on the target lane line point cloud data and perform denoising processing again. Optionally, if there is again a sparse area of the point cloud after denoising processing, encrypted sampling can be performed again for supplementation.

[0130] After performing quality evaluation on the target lane line point cloud data and performing denoising processing again, perform voxel filtering processing on the denoised target lane line point cloud data to generate the final lane line point cloud data; perform vectorization processing on the final lane line point cloud data of the same lane to obtain lane line vectorization data.

[0131] As Figure 7 shown, Figure 7 The figure is a functional module diagram of a lane line point cloud data supplement system proposed in an embodiment of the present invention. Combining with the previous embodiment, the lane line point cloud data supplement system includes a data acquisition module, a clustering analysis module, a denoising and filtering module, a fitting module, a sampling and supplement module, and a vectorization module.

[0132] Among them, the data acquisition module is used to acquire lane line point cloud data, extract information such as the acquisition time and acquisition vehicle of the lane line point cloud data, and perform attribute configuration, including data quality, availability, etc.; the clustering analysis module is used to perform spatial clustering on the lane line point cloud data to obtain the lane line point cloud data of the same lane line, and can also perform ID configuration on the lane line point cloud data of the same lane line and number the lane line point cloud data of the same lane line; the denoising and filtering module is used to perform denoising processing on the lane line point cloud data, statistical filtering denoising processing, and voxel filtering processing on the target lane line point cloud data; the fitting module is used to perform surface fitting on the lane line point cloud data by using the least squares method to obtain a fitting surface and curve fitting to obtain a fitting curve; the sampling and supplement module is used to uniformly sample the fitting curve and use the sliding window method to determine whether there is a point cloud sparse area. If so, perform encrypted sampling on the point cloud sparse area to obtain encrypted sampling points. Finally, calculate the three-dimensional encrypted sampling points by using the fitting surface and place the three-dimensional encrypted sampling points into the lane line point cloud data to obtain the target lane line point cloud data; the vectorization module is used to vectorize the target lane line point cloud data of the same lane line to obtain lane line vectorized data.

[0133] Based on the same inventive concept, another embodiment of the present invention provides a lane line point cloud data supplement device, as Figure 8 shown. Figure 8 is a module block diagram of a lane line point cloud data supplement device proposed in an embodiment of the present invention. The device 400 includes:

[0134] The first acquisition module 401 is used to obtain a fitting curve of the lane line point cloud data by fitting the acquired lane line point cloud data;

[0135] The second determination module 402 is used to determine a point cloud sparse area of the lane line point cloud data according to the lane line point cloud data;

[0136] The third determination module 403 is used to determine a target curve segment on the fitting curve that coincides with the point cloud sparse area according to the fitting curve and the point cloud sparse area;

[0137] The fourth acquisition module 404 is used to obtain encrypted sampling points in the target curve segment by performing encrypted sampling on the target curve segment;

[0138] The fifth acquisition module 405 is used to perform dimensional transformation on the encrypted sampling points through the lane line point cloud data to obtain three-dimensional encrypted sampling points;

[0139] The sixth acquisition module 406 is used to place the three-dimensional encrypted sampling points into the lane line point cloud data to obtain the target lane line point cloud data.

[0140] Optionally, the second determination module 402 includes:

[0141] A first determination sub-module 4021, configured to determine an external circumscribing region of the lane line point cloud data according to the lane line point cloud data;

[0142] A second division sub-module 4022, configured to divide the external circumscribing region into a plurality of sub-regions according to a set step size;

[0143] A third determination sub-module 4023, configured to determine a point cloud sparse region in the plurality of sub-regions according to a point cloud density threshold.

[0144] Optionally, the third determination sub-module 4023 includes:

[0145] A fourth determination sub-module 40231, configured to determine the number of point clouds in each of the plurality of sub-regions according to the lane line point cloud data;

[0146] A fifth determination sub-module 40232, configured to determine a point cloud sparse region in the plurality of sub-regions according to the number of point clouds in each of the plurality of sub-regions and the point cloud density threshold.

[0147] Optionally, the apparatus further includes:

[0148] A seventh determination module 407, configured to determine the number of point clouds in the external circumscribing region according to the lane line point cloud data;

[0149] An eighth determination module 408, configured to determine the point cloud density threshold according to the external circumscribing region and the number of point clouds.

[0150] Optionally, the fourth acquisition module 404 includes:

[0151] A sixth determination sub-module 4041, configured to determine the point cloud density of the point cloud sparse region according to the point cloud sparse region;

[0152] A seventh determination sub-module 4042, configured to determine the number of sampling times for encrypting the sampling of the target curve segment according to the point cloud density and the point cloud density threshold;

[0153] An eighth acquisition sub-module 4043, configured to perform encrypted sampling of the target curve segment for the number of sampling times according to the number of sampling times, and obtain encrypted sampling points in the target curve segment.

[0154] Optionally, the fifth acquisition module 405 includes:

[0155] A ninth acquisition sub-module 4051, configured to obtain a fitting surface of the lane line point cloud data by fitting the acquired lane line point cloud data;

[0156] The tenth obtaining sub-module 4052 is configured to substitute the coordinates of the encrypted sampling points into the surface equation of the fitting surface to obtain three-dimensional encrypted sampling points.

[0157] Optionally, the apparatus further includes:

[0158] The ninth determining module 409 is configured to obtain multiple groups of target lane line point cloud data of the same lane line and determine them as lane line cluster point cloud data;

[0159] The tenth obtaining module 410 is configured to obtain lane line vectorized data by vectorizing the lane line cluster point cloud data.

[0160] Based on the same inventive concept, another embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. The electronic device executes the computer program to implement the lane line point cloud data supplement method according to any one of the first aspects of the embodiments of the present invention.

[0161] Based on the same inventive concept, another embodiment of the present invention provides a vehicle, where the vehicle includes an automatic driving module, and the automatic driving module is configured to obtain lane line point cloud data, and the lane line point cloud data is determined by the lane line point cloud data supplement method according to any one of the first aspects of the embodiments of the present invention.

[0162] For the apparatus embodiment, since it is basically similar to the method embodiment, the description is relatively simple. For the relevant parts, refer to the partial description of the method embodiment.

[0163] Each embodiment in this specification is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. For the same and similar parts among the embodiments, refer to each other.

[0164] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, apparatuses, electronic devices, storage media, or computer program products. Therefore, the embodiments of the present invention can take the form of completely hardware embodiments, completely software embodiments, or embodiments combining software and hardware aspects. Moreover, the embodiments of the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0165] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and purposes of the present invention. The scope of the present invention is defined by the claims and their equivalents.

[0166] Finally, it should also be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or terminal device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or terminal device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or terminal device comprising the said element.

[0167] The above has introduced in detail a lane line point cloud data supplement method, device, electronic device and vehicle provided by the present invention. Specific examples are used in this text to elaborate on the principle and implementation manner of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present application. The above embodiments are only preferred embodiments given to fully illustrate the present invention, and the protection scope of the present invention is not limited thereto. Equivalent substitutions or transformations made by those skilled in the art on the basis of the present invention are within the protection scope of the present invention.

Claims

1. A method for supplementing lane line point cloud data, characterized in that, The method includes: Obtaining a fitting curve of the lane line point cloud data by fitting the acquired lane line point cloud data; Determining a point cloud sparse region of the lane line point cloud data according to the lane line point cloud data; Determining a target curve segment on the fitting curve that coincides with the point cloud sparse region according to the fitting curve and the point cloud sparse region; Obtaining encrypted sampling points in the target curve segment by performing encrypted sampling on the target curve segment; Performing dimensional transformation on the encrypted sampling points through the lane line point cloud data to obtain three-dimensional encrypted sampling points; Placing the three-dimensional encrypted sampling points into the lane line point cloud data to obtain target lane line point cloud data.

2. The method according to claim 1, wherein The determining, according to the lane line point cloud data, the point cloud sparse region of the lane line point cloud data includes: Determining an external region of the lane line point cloud data according to the lane line point cloud data; Dividing the external region into multiple sub-regions according to a set step size; Determining the point cloud sparse region in the multiple sub-regions according to a point cloud density threshold.

3. The method according to claim 2, characterized in that, The determining, according to the point cloud density threshold, the point cloud sparse region in the multiple sub-regions includes: Determining the number of point clouds in each of the multiple sub-regions according to the lane line point cloud data; Determining the point cloud sparse region in the multiple sub-regions according to the number of point clouds in each of the multiple sub-regions and the point cloud density threshold.

4. The method according to claim 2, wherein The determination of the point cloud density threshold includes: Determining the number of point clouds in the external region according to the lane line point cloud data; Determining the point cloud density threshold according to the external region and the number of point clouds.

5. The method according to claim 2, wherein The obtaining, by performing encrypted sampling on the target curve segment, the encrypted sampling points in the target curve segment includes: Determining the point cloud density of the point cloud sparse region according to the point cloud sparse region; Determining the number of sampling times for performing encrypted sampling on the target curve segment according to the point cloud density and the point cloud density threshold; Performing encrypted sampling on the target curve segment for the number of sampling times according to the number of sampling times to obtain the encrypted sampling points in the target curve segment.

6. The method according to claim 1, characterized in that, The performing, through the lane line point cloud data, dimensional transformation on the encrypted sampling points to obtain three-dimensional encrypted sampling points includes: Obtaining a fitting surface of the lane line point cloud data by fitting the acquired lane line point cloud data; Substituting the coordinates of the encrypted sampling points into the surface equation of the fitting surface to obtain three-dimensional encrypted sampling points.

7. The method according to claim 1, characterized in that, The method further includes: Obtaining multiple groups of target lane line point cloud data of the same lane line and determining them as lane line cluster point cloud data; Obtaining lane line vectorized data by vectorizing the lane line cluster point cloud data.

8. A lane line point cloud data supplement device, characterized in that, The device includes: A first obtaining module, configured to obtain a fitting curve of the lane line point cloud data by fitting the acquired lane line point cloud data; A second determining module, configured to determine a point cloud sparse region of the lane line point cloud data according to the lane line point cloud data; A third determining module, configured to determine a target curve segment on the fitting curve that coincides with the point cloud sparse region according to the fitting curve and the point cloud sparse region; A fourth acquisition module, configured to obtain encrypted sampling points in the target curve segment by performing encrypted sampling on the target curve segment; A fifth acquisition module, configured to perform dimensional transformation on the encrypted sampling points through the lane line point cloud data to obtain three-dimensional encrypted sampling points; A sixth acquisition module, configured to place the three-dimensional encrypted sampling points into the lane line point cloud data to obtain target lane line point cloud data.

9. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The electronic device executes the computer program to implement the lane line point cloud data supplement method according to any one of claims 1 to 7.

10. A vehicle, characterized in that, The vehicle includes an automatic driving module, and the automatic driving module is configured to obtain lane line point cloud data, and the lane line point cloud data is determined by the lane line point cloud data supplement method according to any one of claims 1 to 7.

Images