A method for filtering a gradual encryption triangulation point cloud based on slope compensation

Abstract

This invention discloses a progressively denser triangular mesh point cloud filtering method based on slope compensation, comprising: denoising the slope lidar point cloud, removing outliers, and downsampling; segmenting the downsampling point cloud to obtain the slope inclined surface point cloud and the slope horizontal surface point cloud; segmenting the slope inclined surface point cloud to obtain the slope point cloud; and using a progressively denser triangular mesh with slope compensation to filter the rotated slope point cloud and the slope horizontal surface point cloud. This invention employs a coordinate transformation method for high-slope slopes, which can effectively eliminate their slope information, enabling the correct ground minimum point to be found during the filtering process, thereby completing the slope filtering.

Description

Technical Field

[0001] This invention relates to the technical field of mine slope measurement, and in particular to a progressive densification triangular network point cloud filtering method based on slope compensation. Background Technology

[0002] LiDAR (Light Detection and Ranging) technology, with its fully automated, non-contact, and high-precision 3D data acquisition capabilities, has been widely applied in various fields such as land resources, energy, agriculture, forestry, cultural relic protection, power line inspection, intelligent transportation, digital cities, and emergency disaster relief. Point cloud filtering, which aims to automatically filter out ground points and non-ground points from point clouds, is a key aspect of data processing.

[0003] Numerous studies have been conducted both domestically and internationally on point cloud filtering, proposing various point cloud filtering algorithms, such as morphological filtering, slope filtering, progressively encrypted triangular mesh filtering, and quadratic surface iterative filtering. These methods determine ground points by finding the lowest elevation point within a certain neighborhood, but they lack applicability to scenarios with significant terrain undulations, such as steep slopes. The progressively encrypted triangular mesh filtering algorithm, initially proposed by Axelsson, has been refined and improved over many years by researchers, focusing on two key strategies: initial seed point selection and optimized iterative encryption. This has resulted in good filtering performance in most scenarios, making it particularly suitable for applications with undulating terrain. However, it involves numerous parameter settings, leading to high time costs when processing large amounts of point cloud data. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] Therefore, this invention provides a progressive densification triangular network point cloud filtering method based on slope compensation, which can eliminate the filtering inaccuracy problem caused by the slope of the point cloud.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: including denoising the slope lidar point cloud, removing outliers, and downsampling; segmenting the downsampling point cloud to obtain the slope inclined surface point cloud and the slope horizontal surface point cloud; segmenting the slope inclined surface point cloud to obtain the slope point cloud; and filtering the rotated slope point cloud and the slope horizontal surface point cloud using a slope-compensated progressively denser triangular mesh.

[0008] As a preferred embodiment of the progressive densification triangular network point cloud filtering method based on slope compensation described in this invention, the method of removing outliers includes setting a threshold using an elevation histogram to remove the outliers.

[0009] As a preferred embodiment of the progressive densification triangular network point cloud filtering method based on slope compensation described in this invention, the downsampling includes downsampling the slope lidar point cloud by setting a minimum distance threshold between adjacent point clouds.

[0010] As a preferred embodiment of the progressive densification triangular network point cloud filtering method based on slope compensation described in this invention, the segmented processing includes performing least squares curve fitting on the slope bottom and slope top clouds of the slope slope surface according to the following formula:

[0011]

[0012] By setting a curvature threshold, when the curvature is greater than the threshold, a line is drawn connecting the points of maximum curvature at the corresponding slope bottom and top to obtain a segmented point cloud; the segmented point cloud is then fitted until the curvature of the fitted curve is less than the threshold to obtain the slope point cloud; where x i y i Let i = 1, 2, 3…m, where i is the x-coordinate of the curve fitting point. For the fitted curve, This is the sum of squared deviations.

[0013] As a preferred embodiment of the progressive densification triangular mesh point cloud filtering method based on slope compensation described in this invention, the rotating slope point cloud includes: gridding the slope point cloud, setting four meshes, and arbitrarily selecting three lowest points P1, P2, and P3 in each mesh to fit the plane β.

[0014] Calculate the normal vector of the fitted plane β Rotate the fitting plane β such that the rotated fitting plane β is parallel to the xoy plane, and the normal vector of the rotated fitting plane β... Towards the positive Z-axis; according to the normal vector The rotation angle θ is calculated using the dot product; the normal vector is calculated as follows. The cross product of the vectors yields a vector that is simultaneously perpendicular to the normal vector. Rotation axis Based on the rotation angle θ and the rotation axis Obtain the rotation matrix M, and use the rotation matrix M to rotate the slope point cloud.

[0015] As a preferred embodiment of the progressive densification triangular mesh point cloud filtering method based on slope compensation described in this invention, wherein: the normal vector include,

[0016]

[0017]

[0018] Where a, b, and c are the coefficients of the fitted plane β.

[0019] As a preferred embodiment of the progressively encrypted triangular mesh point cloud filtering method based on slope compensation described in this invention, the rotation matrix M includes,

[0020]

[0021] in, For the axis of rotation The unit vector.

[0022] As a preferred embodiment of the slope-compensated progressive densified triangular mesh point cloud filtering method of the present invention, the filtering includes: first, rasterizing the rotated slope point cloud before filtering and recording the lowest point in each grid; performing an opening operation on the raster data to obtain initial ground seed points; constructing an initial triangular mesh using the initial ground seed points; constructing the slope-compensated progressive densified triangular mesh based on the initial triangular mesh; and classifying ground points by judging the iteration angle and iteration height; wherein, the iteration angle is the angle between a point inside a triangle and the vertex of the triangle and the plane formed by the triangle, and the iteration height is the vertical distance from a point inside a triangle to the plane formed by the triangle; the triangle is any triangle in the initial triangular mesh.

[0023] As a preferred embodiment of the progressive densification triangular network point cloud filtering method based on slope compensation described in this invention, the method further includes: converting the rotated slope point cloud into an initial coordinate system using an inverse transformation rotation matrix M1, and fusing the segmented point cloud and the slope horizontal plane point cloud to obtain the final ground point cloud; wherein, the inverse transformation rotation matrix M1 is:

[0024]

[0025] in, The rotation axis of the inverse transformation matrix. The identity matrix is

[0026] The beneficial effects of this invention are as follows: This invention uses a coordinate transformation method for steep slopes, which can effectively eliminate their slope information, so that the correct lowest point on the ground can be found during the filtering process, thereby completing the filtering of the slope. Attached Figure Description

[0027] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0028] Figure 1 This is a flowchart illustrating the progressive densification triangular mesh point cloud filtering method based on slope compensation described in the first embodiment of the present invention.

[0029] Figure 2 This is a schematic diagram of point cloud segmentation on a slope inclined surface using the progressive densification triangular mesh point cloud filtering method based on slope compensation described in the first embodiment of the present invention.

[0030] Figure 3 This is a schematic diagram of slope point cloud rotation for the progressive densification triangular mesh point cloud filtering method based on slope compensation described in the first embodiment of the present invention.

[0031] Figure 4 This is a schematic diagram illustrating the principle of the progressively encrypted triangular mesh filtering algorithm for the point cloud filtering method based on slope compensation described in the first embodiment of the present invention. Detailed Implementation

[0032] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0033] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0034] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0035] This invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not adhering to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0036] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0037] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0038] Example 1

[0039] Reference Figures 1-4 This is the first embodiment of the present invention. This embodiment addresses the characteristics of steep slopes, such as those in open-pit mines, by providing a progressively denser triangular mesh point cloud filtering method based on slope compensation, comprising:

[0040] S1: Denoise the lidar point cloud of the slope, remove outliers, and downsample.

[0041] Thresholds were set using an elevation histogram to remove outliers.

[0042] Furthermore, by setting a minimum distance threshold between adjacent point clouds, the slope lidar point cloud is downsampled to obtain point cloud data with ideal density, thereby reducing data redundancy.

[0043] S2: Segment the downsampled point cloud to obtain the point cloud of the slope inclined surface and the point cloud of the slope horizontal surface.

[0044] This embodiment uses a region growing method to segment the point cloud, obtaining point clouds of the slope inclined surface and point clouds of the slope horizontal surface.

[0045] S3: Segment the point cloud of the slope slope to obtain the slope point cloud.

[0046] The steps for segmentation are as follows:

[0047] (1) According to the following formula, the slope bottom and slope top of the slope surface are respectively fitted by the least squares method to obtain the curve XY equation:

[0048]

[0049] Where, x i y i Let i = 1, 2, 3…m, where i is the x-coordinate of the curve fitting point. For the fitted curve, This is the sum of squared deviations.

[0050] (2) By setting a curvature threshold, when the curvature is greater than the curvature threshold, a connection is made at the point with the maximum curvature at the bottom and top of the slope to obtain a dividing line, and then a segmented point cloud is obtained.

[0051] (3) Fit the segmented point cloud until the curvature of the fitted curve is less than the threshold, at which point the segmentation process is complete, resulting in a relatively gentle slope point cloud; for example Figure 2 As shown.

[0052] S4: Filter the rotating slope point cloud and the slope horizontal plane point cloud using a progressively denser triangular mesh with slope compensation.

[0053] The specific steps for rotating slope point cloud are as follows:

[0054] (1)Reference Figure 3 The slope point cloud is gridded, with four grids set. The lowest point in each grid is taken to obtain four points. Three lowest points P1, P2, and P3 in each grid are arbitrarily selected to fit the plane β, as shown in the following formula:

[0055] β:ax+by+cz+d=0

[0056] Where a, b, c, and d are the coefficients of the fitted plane β.

[0057] (2) Calculate the normal vector of the fitting plane β. Rotate the fitting plane β such that the rotated fitting plane β is parallel to the xoy plane. Let the rotated fitting plane β be the O-XY plane, and the normal vector of this plane be... Towards the positive Z-axis;

[0058] normal vector for:

[0059]

[0060]

[0061] (3) Based on the normal vector The dot product is used to calculate the rotation angle θ.

[0062]

[0063] (4) Calculate the normal vector The cross product of the vectors yields a vector that is simultaneously perpendicular to the normal vector. Rotation axis

[0064] Rotation axis:

[0065]

[0066]

[0067] The identity matrix is ​​denoted as

[0068] (5) Based on the rotation angle θ and the rotation axis Obtain the rotation matrix M, and use the rotation matrix M to rotate the slope point cloud to eliminate its slope influence and obtain the rotated slope point cloud.

[0069] Rotation matrix M:

[0070]

[0071] Furthermore, the rotating slope point cloud and the slope horizontal plane point cloud are filtered using a progressively denser triangular mesh with slope compensation. The specific steps are as follows:

[0072] (1) Before filtering, the rotating slope point cloud is rasterized and the lowest point in each grid is recorded.

[0073] (2) Perform an opening operation on the raster data to obtain the initial ground seed points;

[0074] The opening operation involves erosion followed by dilation. Erosion selects the lowest point in the grid, while dilation selects the maximum value from the lowest point in the grid to obtain the initial ground seed point.

[0075] (3) Construct an initial triangulation using initial ground seed points;

[0076] (4) Construct a progressively denser triangulation network with slope compensation based on the initial triangulation network;

[0077] In constructing a progressively denser triangulated network for slope compensation, ground points are classified by determining the iteration angle and iteration height;

[0078] Wherein, the iteration angle is the angle between a point inside the triangle and the plane formed by the vertices of the triangle and the triangle itself, such as... Figure 4 In this context, χ represents the iteration height, which is the perpendicular distance from a point inside the triangle to the plane formed by the triangle, such as... Figure 4 S represents the triangle; a triangle is any triangle in the initial triangulation, constructed from the initial ground seed points;

[0079] (5) The rotated slope point cloud is converted into the initial coordinate system by the inverse transformation rotation matrix M1, and the segmented point cloud and the slope horizontal plane point cloud are fused to obtain the final ground point cloud.

[0080] The inverse transformation rotation matrix M1 is:

[0081]

[0082] in, The rotation axis of the inverse transformation matrix. The identity matrix is

[0083] Preferably, this method first segments the slope by analyzing the changes in slope curvature, then grids the segmented point clouds. For each segment, the lowest elevation point in each grid is used for plane fitting. The angle between the fitted plane and the horizontal plane is then calculated, and the fitted plane is rotated to a horizontal plane to eliminate the influence of slope inclination on the filtering results. For the transformed point cloud, a progressively denser triangular network filtering algorithm is used. First, the point cloud is gridded, then an erosion-dilation method is used to select initial ground seed points to construct an irregular triangular network (TIN). Finally, a reasonable iteration height and iteration angle are selected to densify the triangular network, achieving the effect of slope filtering.

[0084] Example 2

[0085] To verify and illustrate the technical effectiveness of this method, this embodiment selects traditional technical solutions (slope filtering, TPS filtering) and compares them with this method for testing. The test results are compared using scientific methods to verify the real effectiveness of this method.

[0086] Traditional techniques generally have poor filtering performance for near-ground points and are time-consuming to process large amounts of point cloud data. To verify that this method has better filtering performance and processing efficiency than traditional techniques, this embodiment will use both traditional techniques and this method to filter point cloud data.

[0087] In this embodiment, the SIte2 reference data published online by ISPRS was selected as the experimental data. Each point in the experimental area was marked as a ground point or a non-ground point. The experimental area covered an area of ​​630m × 420m and contained 243,400 points. The area contained complex buildings, large buildings, and data gaps, with an average point spacing of 1 to 1.5m.

[0088] To quantitatively analyze the experimental results, this embodiment uses two types of errors for evaluation: the filtering quality can be represented by Type I error (i.e., the error of ground points being misclassified as ground feature points) and Type II error (i.e., the error of ground feature points being treated as ground points); the calculation method for the errors is as follows:

[0089] Type 1 error:

[0090]

[0091] Type 2 error:

[0092]

[0093] Total error:

[0094]

[0095] Where a represents correctly classified ground points, b represents ground points incorrectly classified as ground features, c represents ground features incorrectly classified as ground points, and d represents correctly classified ground features. The filtered point cloud data is statistically analyzed and compared with the standard classification results given by the reference data, as shown in Table 1.

[0096] Table 1: Analysis of filtering error results.

[0097]

[0098] As can be seen from the table above, this method filters ground points and non-ground points with different thresholds, which effectively controls the possibility of low-lying points being incorrectly classified as ground points. Furthermore, the Type II error and the total error are reduced compared to the slope method and TPS method.

[0099] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A slope-compensation-based, progressively-encrypted, TIN-cloud filtering method, characterized in that: The method comprises the following steps: The slope laser radar point cloud is denoised, abnormal points are removed, and downsampling is performed; The downsampling point cloud is segmented to obtain slope inclined surface point cloud and slope horizontal surface point cloud; The slope inclined surface point cloud is processed in segments to obtain slope point cloud; The rotated slope point cloud and the slope horizontal surface point cloud are filtered by using the slope-compensated progressive densification triangulation network. The rotated slope point cloud comprises, The slope point cloud is gridded, four grids are set, and three lowest points in each grid are randomly selected , , to fit a plane ; Calculate the fitted plane normal vector Rotate the fitting plane This makes the fitted plane of the rotation... Parallel to the xoy plane, and the fitted plane after rotation normal vector Towards the positive Z-axis; Based on the normal vector , The dot product is used to calculate the rotation angle θ. By calculating the cross product of the normal vector , , the rotation axis is obtained which is simultaneously perpendicular to the normal vector , . According to the rotation angle θ and the rotation axis A rotation matrix M is obtained, and the slope point cloud is rotated using the rotation matrix M.

2. The slope-compensation-based, CIE Triangular Gird Point Cloud filtering method of claim 1, wherein: The removal of abnormal points comprises, The threshold is set by using the elevation histogram to remove the abnormal points.

3. The slope-compensation-based, CEGT cloud filtering method of claim 1, wherein: The downsampling comprises, The slope laser radar point cloud is subjected to the downsampling by setting a minimum distance threshold of adjacent point clouds.

4. The slope-compensated, CEGT cloud filtering method of claim 2 or 3, wherein: The segment processing comprises, The slope bottom and slope top point clouds of the slope inclined surface are respectively subjected to least square curve fitting according to the following formula: ； By setting a curvature threshold, when the curvature is greater than the curvature threshold, the corresponding slope bottom and slope top curvature maximum points are connected to obtain a segmentation line, and then segmented point cloud is obtained; The segmented point cloud is fitted until the curvature of the fitted curve is less than the threshold, and the slope point cloud is obtained; wherein, Xi, yi are the horizontal and vertical coordinates of the curve fitting points, respectively, i = 1, 2, 3…m; is the fitting curve, is the sum of squared deviations.

5. The slope-compensation-based, CEGT cloud filtering method of claim 1, wherein: The normal vector , comprises, ； ； where a, b, c are coefficients of the fitting plane .

6. The slope-compensation-based, CEGT cloud filtering method of claim 5, wherein: The rotation matrix M comprises, ； wherein is a unit vector of the rotation axis of the rotation axis.

7. The slope-compensation-based, CEGT cloud filtering method of claim 1, wherein: The filtering comprises, Before filtering, the rotated slope point cloud is rasterized, and the lowest point in each grid is recorded; An initial ground seed point is obtained by performing an open operation on the grid data; An initial triangulation network is constructed by using the initial ground seed point; The slope-compensated progressive densification triangulation network is constructed based on the initial triangulation network, and the ground points are classified by judging an iteration angle and an iteration height; the iteration angle is the angle between a point in a triangle and the vertex of the triangle and the plane formed by the triangle, and the iteration height is the vertical distance from the point in the triangle to the plane formed by the triangle; the triangle is any one of the triangles in the initial triangulation network.

8. The slope-compensated, CEGT cloud filtering method of claim 4, wherein: The method further comprises the following steps: by inverse transforming the rotation matrix convert the rotated slope point cloud into the initial coordinate system, and fuse the segmented point cloud and the slope horizontal plane point cloud to obtain a final ground point cloud; wherein the inverse rotation matrix is: ； wherein is the inverse transform matrix rotated by 90 degrees, is the identity matrix .

Images