A method and apparatus for road boundary extraction based on laser point clouds

Abstract

This invention belongs to the field of high-precision map production technology. It provides a method and apparatus for road boundary extraction based on laser point clouds. The method includes: acquiring laser point cloud data; coarsely extracting road boundary points from the laser point cloud data; performing a perspective transformation on the coarsely extracted road boundary points; finely extracting the perspective-transformed road boundary points; and fitting the finely extracted road boundary points to obtain the road boundary line. This invention, after coarsely extracting road boundary points, performs a perspective transformation on the road boundary points, making the originally sparse road boundary points more compact, effectively enhancing the linear characteristics of the real road boundary points, reducing the influence of interference points with similar features, and thus improving the accuracy of the road boundary extraction algorithm.

Description

Technical Field

[0001] This invention belongs to the field of high-precision map production technology, and in particular relates to a method and apparatus for extracting road boundaries based on laser point clouds. Background Technology

[0002] Road boundary lines are one of the key road elements that distinguish road areas from their surroundings. Laser point clouds are a major data source for high-precision maps. They are collections of massive points obtained using laser scanning technology, which can accurately represent the location distribution and surface features of objects in three-dimensional space, providing a data foundation for the rapid acquisition and updating of road data.

[0003] Current research on road boundary line extraction based on laser point clouds can be mainly divided into three categories: based on two-dimensional feature images of point clouds, based on spatial distribution features of scan lines, and based on three-dimensional morphological features of point clouds. These methods, in the process of extracting road boundary points, either rely heavily on manual thresholding or are susceptible to interference points, leading to limited accuracy and insufficient refinement of the generated road boundary lines. Furthermore, the extracted road boundary points are often sparse, and there are interference points with similar features, resulting in inaccurate road boundary lines ultimately fitted.

[0004] Therefore, overcoming the shortcomings of the existing technology is an urgent problem to be solved in this technical field. Summary of the Invention

[0005] This invention provides a solution to the technical problem that existing methods extract sparse road boundary points and have interference points with similar characteristics, which leads to inaccurate fitted road boundary lines.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a road boundary extraction method based on laser point clouds, comprising:

[0008] Acquire laser point cloud data, and coarsely extract road boundary points from the laser point cloud data to obtain candidate road boundary points;

[0009] The road boundary candidate points are transformed by a viewpoint to obtain the road boundary candidate points after the viewpoint transformation.

[0010] The candidate points of the road boundary after the viewpoint transformation are refined and fitted to obtain the road boundary line.

[0011] Preferably, the coarse extraction of road boundary points from the laser point cloud data to obtain candidate road boundary points includes:

[0012] The road point cloud is laterally segmented based on the trajectory data sampled at intervals to obtain a road cross-section sequence;

[0013] The road section sequence is divided into grids, the principal points of the grid cells are calculated, and pseudo-scan lines are generated.

[0014] Calculate the slope and elevation difference between two adjacent points on the pseudo-scan line. Points that are within both the slope threshold and elevation difference threshold are identified as candidate points for the road boundary.

[0015] Preferably, the trajectory data based on interval sampling is used to laterally segment the road point cloud to obtain a road cross-section sequence, including:

[0016] Acquire trajectory data of the acquisition device when collecting laser point clouds, and sample trajectory points in the trajectory data at intervals at certain distances;

[0017] Centered on the sampled trajectory points, the interception width is set, the road point cloud is divided into horizontal blocks, and the intercepted road point cloud is projected onto the corresponding cross section to obtain the road cross section sequence.

[0018] Preferably, the step of dividing the road cross-section sequence into a grid, calculating the principal points of the grid cells, and generating pseudo-scan lines includes:

[0019] Grid the series of sections using the set cut width as the unit length;

[0020] Sort all points in each grid cell by elevation, calculate the elevation difference between adjacent points, divide points with elevation difference less than the threshold into the same layer of data, and take the highest point of the bottom layer of data as the master point of the grid cell.

[0021] Connect the principal points of each grid cell to generate pseudo-scan lines.

[0022] Preferably, the step of performing a perspective transformation on the road boundary candidate points to obtain the perspective-transformed road boundary candidate points includes:

[0023] Obtain the indices of candidate points for road boundaries; where the indices correspond to the point cloud coordinate values.

[0024] Calculate the center point of the point cloud for the current road segment based on the maximum and minimum point cloud coordinates of the current road segment;

[0025] Calculate the mean heading angle of the trajectory points, and the vertical direction of the mean heading angle of the trajectory points from above, and obtain the perpendicular line of the mean heading angle passing through the center point of the point cloud;

[0026] Rotate the road boundary point cloud around the perpendicular line of the mean heading angle, and project the rotated road boundary point cloud onto the xoy plane to obtain the road boundary candidate points after the view transformation.

[0027] Preferably, the rotation of the road boundary point cloud is performed in a clockwise or counterclockwise manner.

[0028] Preferably, the clockwise or counterclockwise rotation angle is 15°-75°.

[0029] Preferably, the step of refining and fitting the candidate points of the road boundary after the viewpoint transformation to obtain the road boundary line includes:

[0030] The candidate points of the road boundary after the viewpoint change are subjected to initial curve fitting. Points outside the tolerance range in the fitting results are removed as outliers of the non-road boundary line. Points within the tolerance range in the fitting results are used as the optimized road boundary points.

[0031] Based on the original index of the fitted and optimized road boundary points, find the candidate road boundary points before the view transformation, and use them as the finely extracted road boundary points.

[0032] The extracted road boundary points are then subjected to curve fitting again. Any missing points in the fitting results are filled by interpolation to obtain the final road boundary line.

[0033] Preferably, the initial curve fitting of the candidate road boundary points after the viewpoint change is performed using the random sampling consistency method.

[0034] Secondly, the present invention provides a road boundary extraction device based on laser point clouds, comprising:

[0035] At least one processor; and,

[0036] A memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the processor for performing the road boundary extraction method based on laser point clouds as described in the first aspect.

[0037] In view of the shortcomings of the prior art, the beneficial effects that the present invention can achieve are as follows:

[0038] This invention proposes a road boundary extraction method and apparatus based on laser point clouds. After coarse extraction of road boundary points, the method transforms the perspective of the road boundary points to make the originally sparse road boundary points more compact, effectively enhancing the linear features of the real road boundary points, reducing the influence of interference points with similar features, and thus improving the accuracy of the road boundary extraction algorithm. Attached Figure Description

[0039] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments of the present invention will be briefly described below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0040] Figure 1 This is a schematic diagram of the process of a road boundary extraction method based on laser point clouds provided in Implementation 1;

[0041] Figures 2-6 yes Figure 1 A flowchart illustrating each step in the process;

[0042] Figure 7 This is a schematic diagram illustrating the principle of a road boundary extraction method based on laser point clouds provided in Implementation 1;

[0043] Figure 8 This is a data processing diagram of a road boundary extraction method based on laser point clouds provided in Implementation 1;

[0044] Figure 9 This is a schematic diagram of a road boundary extraction device based on laser point clouds provided in Implementation 2. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. In addition, the technical features of the various embodiments or individual embodiments provided by the present invention can be arbitrarily combined with each other to form feasible technical solutions. Such combinations are not constrained by the order of steps and / or structural composition patterns, but must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0046] Example 1:

[0047] To address the technical problems of sparse road boundary points extracted by existing methods and the presence of interference points with similar features, leading to inaccurate fitted road boundary lines, this embodiment 1 provides a road boundary extraction method based on laser point clouds, such as... Figure 1 As shown, it includes:

[0048] S100: Acquire laser point cloud data, perform coarse extraction of road boundary points in the laser point cloud data, and obtain candidate road boundary points.

[0049] In this step, the laser point cloud data is a set of vectors in a three-dimensional coordinate system acquired by a laser point cloud acquisition device. Commonly used laser point cloud acquisition devices, such as three-dimensional laser scanners, can be divided into short-range, medium-range, and long-range according to their acquisition distance.

[0050] As one implementation method, the road boundary points in the laser point cloud data are coarsely extracted to obtain candidate road boundary points, such as... Figure 2 As shown, it includes:

[0051] S110, based on the trajectory data sampled at intervals, the road point cloud is laterally segmented to obtain the road cross-section sequence.

[0052] In specific implementation, the trajectory data based on interval sampling is used to laterally segment the road point cloud to obtain a road cross-section sequence, such as... Figure 3 As shown, it includes:

[0053] S111: Acquire the trajectory data of the acquisition device when collecting laser point clouds, and sample the trajectory points in the trajectory data at intervals at a certain distance.

[0054] In the application process, in order to make the point cloud data collected by the laser point cloud acquisition device more comprehensive and accurate, it is usually necessary to scan the road segment in multiple directions according to a predetermined path. During the scanning process, the trajectory data of the acquisition device is generated. After acquiring the trajectory data of the acquisition device, trajectory data points are collected at certain intervals r (r can be estimated in advance based on the road segment conditions and map accuracy). The trajectory data of the acquisition device is then segmented based on the trajectory data points.

[0055] S112, with the sampled trajectory points as the center, set the interception width, divide the road point cloud into horizontal blocks, and project the intercepted road point cloud onto the corresponding cross section to obtain the road cross section sequence.

[0056] During application, the above-sampled trajectory points pos i Centered on (i = 1, ..., n), with a cut-off width of c, the road point cloud is horizontally divided into blocks. After division, the cut-off road point cloud is projected onto the corresponding cross-section to obtain the road cross-section sequence Road_Split. i (i = 1, ..., n).

[0057] S120: Divide the road cross-section sequence into a grid, calculate the principal points of the grid cells, and generate pseudo-scan lines.

[0058] In specific implementation, the road cross-section sequence is divided into grids, the principal points of the grid cells are calculated, and pseudo-scan lines are generated, such as... Figure 4 As shown, it includes:

[0059] S121, using the set cut width as the unit length, divide each section series into grids.

[0060] That is, each cross-section series is divided into grids using the cut width c set above as the unit length.

[0061] S122: Sort all points in each grid cell according to elevation, calculate the elevation difference between adjacent points, divide points with elevation difference less than the threshold into the same layer of data, and take the highest point of the bottom layer of data as the master point of the grid cell.

[0062] The elevation refers to the distance from all points within each grid cell along the vertical direction to the absolute datum plane. In application, the height difference (i.e., elevation difference) between adjacent points is set as Δh, and the elevation difference Δh is compared with a preset elevation difference threshold h. t Compare the elevation differences, and assign the elevation difference Δh to the threshold h. t Points are assigned to the same data layer, and the highest point in the lowest data layer is taken as the principal point of the grid cell.

[0063] S123 connects the principal points of each grid cell to generate pseudo-scan lines.

[0064] S130: Calculate the slope and elevation difference between two adjacent points on the pseudo-scan line. Points that are within both the slope threshold and elevation difference threshold are identified as candidate points for the road boundary.

[0065] During the application, let the slope s and elevation difference h be the values ​​of two adjacent points on the pseudo-scan line. Points that fall within both the slope threshold and elevation difference threshold are identified as candidate points for the road boundary, pt. j (j = 1, ..., m).

[0066] S200, the viewpoint of the candidate road boundary is transformed to obtain the candidate road boundary after viewpoint transformation.

[0067] In this step, the distance between the originally sparse road boundary candidate points is made more compact by changing the perspective of the road boundary candidate points.

[0068] As one implementation method, the road boundary candidate points are subjected to view transformation to obtain the road boundary candidate points after view transformation, such as... Figure 5 As shown, it includes:

[0069] S210, obtain the index of the candidate points of the road boundary; where the index corresponds to the point cloud coordinate value.

[0070] During application, the index of the candidate points for the road boundary is set to pt_index. j (j = 1, ..., m).

[0071] S220: Calculate the center point of the point cloud of the current road segment based on the maximum and minimum point cloud coordinates of the current road segment.

[0072] During application, the center point of the point cloud is set as CenterPt.

[0073] S230, calculate the mean heading angle of the trajectory points, and the top vertical direction of the mean heading angle of the trajectory points, and obtain the perpendicular line of the mean heading angle passing through the center point of the point cloud.

[0074] During application, the yaw angle of the trajectory point is set. i The mean yaw_mean of (i=1,…,n) is calculated, the angle of the mean heading angle of the trajectory points is calculated as angle=yaw_mean+90°, and the axis_line of the mean heading angle passing through the center point is obtained.

[0075] S240, rotate the road boundary point cloud with the perpendicular line of the mean heading angle as the axis, and project the rotated road boundary point cloud onto the xoy plane to obtain the road boundary candidate points after the view transformation.

[0076] In practice, the road boundary point cloud is rotated around the axis_line perpendicular to the mean heading angle and projected onto the xoy plane to obtain the road boundary candidate points pt2d after the view transformation. j (j = 1, ..., m).

[0077] Preferably, the road boundary point cloud is rotated in a clockwise or counterclockwise direction, and the rotation angle is 15°-75°. In application, 45° is preferred.

[0078] like Figure 7 The figure shown is a schematic diagram of the principle of a road boundary extraction method based on laser point cloud provided in Implementation 1. Taking a road boundary point cloud as an example, Figure (b) is obtained by rotating Figure (a) by a certain angle in space and projecting it. It can be seen that after rotation and projection, distance D1 < distance D2.

[0079] S300 performs fine extraction and fitting on the candidate points of the road boundary after the viewpoint transformation to obtain the road boundary line.

[0080] As one implementation method, the road boundary candidate points after the viewpoint transformation are refined and fitted to obtain the road boundary line, such as... Figure 6 As shown, it includes:

[0081] S310, perform initial curve fitting on the candidate points of the road boundary after the viewpoint transformation, remove the points outside the tolerance range in the fitting results as outliers of the non-road boundary line, and use the points within the tolerance range in the fitting results as the optimized road boundary points.

[0082] During application, the candidate points for the road boundary after the viewpoint transformation are pt2d. j (j=1,…,m), the outliers are removed, and the optimized road boundary points are boundary_pt2d. j (j=1,…,s); Preferably, the initial curve fitting of the candidate road boundary points after the viewpoint transformation is performed using the random sampling consensus method. By using the random sampling consensus method, the optimal road boundary points can be retained during the outlier removal process, thereby obtaining the optimal road boundary line.

[0083] S320: Based on the original index of the fitted and optimized road boundary points, find the candidate road boundary points before the view transformation, and use them as the refined road boundary points.

[0084] During application, the original index of the road boundary points is pt_index. j (j=1,…,m), by finding the point set before the viewpoint transformation, which is the finely extracted road boundary point boundary_pts.

[0085] S330, perform curve fitting again on the finely extracted road boundary points, and interpolate to fill in any missing points in the fitting results to obtain the final road boundary line.

[0086] During the application process, the extracted road boundary points (boundary_pts) are subjected to curve fitting again to finally obtain the road boundary lines (road_boundary_lines) of the current road segment.

[0087] like Figure 8 The figure shown is a data processing diagram of a road boundary extraction method based on laser point cloud provided in Implementation 1. Through a series of processing of road point cloud data, a more accurate road boundary is obtained.

[0088] This embodiment 1 provides a road boundary extraction method based on laser point clouds. After coarse extraction of road boundary points, the method transforms the perspective of the road boundary points to make the originally sparse road boundary points more compact, effectively enhancing the linear features of the real road boundary points, reducing the influence of interference points with similar features, and thus improving the accuracy of the road boundary extraction algorithm.

[0089] Example 2:

[0090] Based on the same overall technical solution as in Embodiment 1, such as Figure 9 The diagram shown is a schematic of a road boundary extraction device based on laser point clouds provided in Embodiment 2, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the processor to perform the road boundary extraction method based on laser point clouds as described in Embodiment 1.

[0091] In summary, this invention provides a road boundary extraction method and apparatus based on laser point clouds. After coarse extraction of road boundary points, the method transforms the perspective of the road boundary points to make the originally sparse road boundary points more compact, effectively enhancing the linear features of the real road boundary points, reducing the influence of interference points with similar features, and thus improving the accuracy of the road boundary extraction algorithm.

[0092] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0093] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, electronic devices, or computer software program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0094] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, systems, electronic devices, or computer software program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A system that specifies functions in one or more boxes.

[0095] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including an instruction set implemented in a process. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0096] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0097] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

Claims

1. A method for extracting a road boundary based on a laser point cloud, characterized in that, The method comprises the following steps: acquiring laser point cloud data, and roughly extracting road boundary points in the laser point cloud data to obtain road boundary candidate points; performing perspective transformation on the road boundary candidate points to obtain road boundary candidate points after perspective transformation; performing fine extraction and fitting on the road boundary candidate points after perspective transformation to obtain a road boundary line; the perspective transformation on the road boundary candidate points to obtain road boundary candidate points after perspective transformation comprises: acquiring an index of the road boundary candidate points; wherein the index corresponds to a point cloud coordinate value; calculating a point cloud center point of a current road section according to a maximum value and a minimum value of the point cloud coordinates of the current road section; calculating a mean value of the heading angles of the trajectory points, and a vertical line passing through the point cloud center point in the mean value of the heading angles in the vertical direction of the top view to obtain a mean value heading angle vertical line; rotating the road boundary point cloud about the mean value heading angle vertical line as an axis, and projecting the rotated road boundary point cloud onto an x-o-y plane to obtain road boundary candidate points after perspective transformation. 2.The method of claim 1, wherein, the rough extraction of the road boundary points in the laser point cloud data to obtain road boundary candidate points comprises: horizontally cutting the road point cloud based on the trajectory data sampled at intervals to obtain a road section sequence; dividing the road section sequence into grids, calculating the main points of the grid units, and generating pseudo-scan lines; calculating the slope and height difference between any two points of the pseudo-scan lines, and marking the points within the slope threshold and height difference threshold as road boundary candidate points. 3.The method of claim 2, wherein, the horizontal cutting of the road point cloud based on the trajectory data sampled at intervals to obtain a road section sequence comprises: acquiring trajectory data of a collection device when collecting laser point cloud, and sampling the trajectory points in the trajectory data at intervals according to a certain distance; setting a cutting width with the sampled trajectory points as the center, horizontally dividing the road point cloud, and projecting the cut road point cloud onto the corresponding section to obtain a road section sequence. 4.The method of claim 3, wherein, the grid division of the road section sequence, the calculation of the main points of the grid units, and the generation of the pseudo-scan lines comprise: dividing each section series into grids with the set cutting width as the unit length; sorting all points in each grid unit according to the elevation, calculating the height difference between adjacent points, dividing the points with a height difference less than a height difference threshold into the same layer of data, and taking the highest point of the bottom layer of data as the main point of the grid unit; connecting the main points of each grid unit to generate pseudo-scan lines. 5.The method of claim 1, wherein, the rotation of the road boundary point cloud is clockwise rotation or counterclockwise rotation. 6.The method of claim 5, wherein, the clockwise rotation or counterclockwise rotation is at an angle of 15°-75°. 7.The method of claim 1, wherein, the fine extraction and fitting of the road boundary candidate points after perspective transformation to obtain a road boundary line comprises: performing primary curve fitting on the road boundary candidate points after perspective transformation, removing the points outside the tolerance range in the fitting result as outliers of the non-road boundary line, and taking the points within the tolerance range in the fitting result as the road boundary points after fitting optimization; finding the road boundary candidate points before perspective transformation according to the original index of the road boundary points after fitting optimization as the fine extracted road boundary points; The road boundary points extracted are subjected to secondary curve fitting, and missing points in the fitting result are interpolated to obtain the final road boundary line. 8.The method of claim 7, wherein, The road boundary candidate points subjected to the view angle transformation are subjected to primary curve fitting, and the primary curve fitting method adopts a random sample consensus method. 9.A device for extracting a road boundary based on a laser point cloud, characterized by, The method comprises the following steps: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the processor to execute the method for extracting road boundaries based on laser point clouds according to any one of claims 1-8.

Images