A strip point cloud edge joint difference acquisition method and system
By selecting flat areas in strip point clouds and using gridding and KD tree search, the problem of insufficient accuracy and efficiency in obtaining edge difference in existing technologies is solved, achieving high-precision edge difference acquisition with low computational cost, which is applicable to various terrains.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 自然资源部第一航测遥感院(陕西省第五测绘工程院)
- Filing Date
- 2026-04-14
- Publication Date
- 2026-07-10
AI Technical Summary
Existing technologies, when acquiring the edge difference of strip point clouds, are affected by complex terrain and noise, making it difficult to establish a reliable point correspondence, resulting in insufficient accuracy and efficiency, and failing to meet the needs of engineering applications.
By employing a specific flat area selection strategy and multiple threshold constraints, and using grid division and KD tree search, flat grids are selected and edge differences are calculated to eliminate interference from terrain undulations and ensure that the calculations are based on flat areas.
It achieves high-precision, low-computation acquisition of edge difference, is applicable to various terrains, improves the accuracy of splicing quality evaluation and error correction, and solves the problem of large calculation errors in traditional methods.
Smart Images
Figure CN122367969A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and more specifically to a method and system for obtaining the edge difference of strip point cloud. Background Technology
[0002] LiDAR, with its efficient and precise 3D data acquisition capabilities, plays a crucial role in fields such as 3D terrain modeling, coastal zone mapping, urban 3D reconstruction, and environmental perception for autonomous driving. When dealing with large scenes or wide areas, limited by the range of a single measurement, a strip scanning strategy is typically employed. This involves acquiring multiple overlapping point cloud strips through multiple parallel scans, and then relying on stitching and fusion to generate a complete regional point cloud model. Therefore, high-quality stitching of the strip point clouds is fundamental to ensuring the geometric accuracy and reliability of the final 3D model, and is a key link supporting the aforementioned engineering applications in multiple fields.
[0003] In the process of strip point cloud stitching, the edge difference is a core indicator for evaluating stitching quality and correcting systematic errors. Accurately obtaining the edge difference not only objectively verifies the effectiveness and accuracy of the stitching algorithm, but also provides crucial data support for subsequent error compensation, terrain optimization, and geometric consistency assurance, thereby significantly improving the reliability of terrain analysis and engineering applications. Therefore, accurately obtaining the edge difference is not only an important basis for evaluating stitching quality, but also a fundamental step in promoting the development of point cloud data processing technology towards higher precision and higher reliability.
[0004] Currently, existing technologies for obtaining the edge-to-edge difference of strip point clouds mainly rely on point-by-point comparison methods after point cloud registration. The principle is to quantify the edge-to-edge error by calculating the Euclidean distance or normal deviation between corresponding point pairs within the overlapping area after registering adjacent strips. The theoretical premise of this method is the ability to accurately establish the correspondence between points. However, in practical applications, when lidar performs strip scanning, the massive distribution of the strip point cloud is not uniform due to the influence of complex terrain and noise. Adjacent strips scan the same area from different spatial locations, resulting in a lack of significant geometric features in the overlapping areas of adjacent strips. Feature extraction and matching are difficult, making it hard to establish reliable point correspondences. Therefore, existing edge-to-edge difference acquisition methods are insufficient in terms of accuracy, efficiency, and adaptability to meet the increasingly stringent engineering application requirements. Summary of the Invention
[0005] To address the problems existing in the above-mentioned fields, this invention proposes a method and system for obtaining the edge difference of strip point clouds, which can achieve high-precision extraction of the edge difference of strip point clouds through specific flat area screening strategies and multiple threshold constraints.
[0006] To address the aforementioned technical problems, this invention discloses a method for obtaining the edge difference of strip point clouds, comprising the following steps: Obtain the planar overlap region of the reference strip and the target strip, perform meshing on the planar overlap region, and determine multiple mesh cells and their corresponding mesh center points in the planar overlap region; For each grid cell, when the total number of reference strip point clouds is greater than or equal to the number threshold, the reference strip point cloud is divided into data blocks; the multiple temporary candidate points that are closest to the grid center point of the current grid cell in each point cloud block are obtained; the multiple temporary candidate points corresponding to each point cloud block are merged, and the multiple temporary candidate points that are closest to each other are selected from the merged multiple temporary candidate points as multiple candidate points of the reference strip point cloud; When multiple candidate points are all located within the inscribed circle of the current grid cell and the range is less than the range threshold, the current grid cell is determined to be a flat grid; the point in the reference strip that is closest to the grid center point of the current grid cell is obtained from the flat grid. Using the position of a point in the reference strip as the center, obtain the point in the target strip point cloud that is closest to the point in the reference strip, forming a point pair between the reference strip and the target strip; when the distance between a point in the target strip and a point in the reference strip is less than a distance threshold, the point pair is considered a valid point pair; Based on the valid point pairs, a quantitative index of the point cloud edge difference between the reference strip and the target strip is determined by generating a difference array between the points in the reference strip and the points in the target strip.
[0007] Preferably, for each grid cell, when the total number of reference strip point clouds is greater than or equal to a threshold, the reference strip point cloud is divided into data blocks; multiple temporary candidate points closest to the grid center point of the current grid cell are obtained in each point cloud block; the multiple temporary candidate points corresponding to each point cloud block are merged, and the multiple temporary candidate points closest to the merged temporary candidate points are selected as multiple candidate points of the reference strip point cloud, specifically including: For each grid cell, when the total number of reference strip point clouds is greater than or equal to the number threshold, the reference strip point cloud is divided into data blocks to obtain multiple point cloud blocks. Using each point cloud block as a unit, a KD-tree search is used to obtain the distance of each point in the current point cloud block from the center point of the current grid cell; then, the distances are sorted from smallest to largest, and the points with the highest distances are selected. k A temporary candidate point is used as the corresponding point cloud block. k One temporary candidate site; Correspondingly obtain the other point cloud blocks k One temporary candidate site; The corresponding points in each point cloud block k The temporary candidate points are merged to obtain multiple temporary candidate points after merging. Sort the merged temporary candidate points in ascending order of distance, and select the highest-ranked ones. k A temporary candidate point is used as a reference strip point cloud. k There are 10 candidate points.
[0008] Preferably, when all candidate points are located within the inscribed circle of the current grid cell and the range is less than the range threshold, the current grid cell is determined to be a flat grid; obtaining the point in the reference strip that is closest to the grid center point of the current grid cell from the flat grid specifically includes: When the distance to the last candidate point in the sorting of multiple candidate points in the reference strip point cloud is less than half the grid size of the current grid cell, it is determined that all multiple candidate points are located within the inscribed circle of the current grid cell. Determine the elevation values of multiple candidate points located within the inscribed circle of the current grid cell, and use the difference between the maximum and minimum elevation values as the range of the current grid cell; When the range is less than the range threshold, the current grid cell is determined to be a flat grid. The point in the reference strip that is closest to the center point of the current grid cell in the flat grid is obtained by KD tree search.
[0009] Preferably, the step of obtaining the point in the target strip that is closest to the point in the reference strip from the point cloud of the target strip, using the position of the point in the reference strip as the center, to form a point pair between the reference strip and the target strip, specifically includes: Using the planar position of a point in the reference strip as the center, a KD search is used to obtain the point in the target strip that is closest to the point in the reference strip in the target strip point cloud, forming a point pair between the reference strip and the target strip.
[0010] Preferably, the step of determining the quantitative index of the point cloud edge difference between the reference strip and the target strip by generating a difference array between points in the reference strip and points in the target strip based on valid point pairs specifically includes: Based on the valid point pairs, the difference array D between the points in the reference strip and the points in the target strip is generated as follows: in, Q 1 represents a point in the reference strip. Q 2 represents a point within the target strip; The quantile method is used to remove outliers from the difference array D, and the mean and mean error of the difference array are calculated as quantitative indicators of the point cloud edge difference between the reference strip and the target strip.
[0011] Preferably, for each grid cell, it further includes: When the total number of reference strip point clouds is less than the threshold, a KD-tree search is used to obtain the distance of each point in the reference strip point cloud from the center point of the current grid cell; the distances are sorted from smallest to largest, and the points with the highest distances are selected. k 10 candidate points, used as reference strip point clouds k There are 10 candidate points.
[0012] Preferably, the step of obtaining the planar overlap region of the reference strip and the target strip, dividing the planar overlap region into a grid, and determining multiple grid cells of the planar overlap region and their corresponding grid center points specifically includes: The planar coverage vectors of the reference strip and the target strip are obtained using a point cloud planar coverage method based on grid segmentation. The intersection operation is performed on the planar coverage vectors of the obtained reference strip and target strip to obtain the planar overlapping area vector range of the reference strip and target strip; Based on the vector range of the planar overlapping region, the planar overlapping region of the reference strip and the target strip is meshed to determine multiple mesh cells corresponding to the planar overlapping region and their corresponding mesh center points.
[0013] Preferably, it also includes a strip point cloud edge difference acquisition system, comprising: The mesh generation module is used to obtain the planar overlapping area of the reference strip and the target strip, perform mesh generation on the planar overlapping area, and determine multiple mesh cells and their corresponding mesh center points in the planar overlapping area. The flat grid generation module is used to divide the reference strip point cloud into blocks for each grid cell when the total number of reference strip point clouds is greater than or equal to the number threshold; obtain the multiple temporary candidate points that are closest to the grid center point of the current grid cell in each point cloud block; merge the multiple temporary candidate points corresponding to each point cloud block, and select the multiple temporary candidate points that are closest to each other from the merged multiple temporary candidate points as multiple candidate points of the reference strip point cloud; The valid point pair generation module is used to determine that the current grid cell is a flat grid when multiple candidate points are all located within the inscribed circle of the current grid cell and the range is less than the range threshold; to obtain the point in the reference strip that is closest to the grid center point of the current grid cell; to obtain the point in the target strip point cloud that is closest to the point in the reference strip, using the position of the point in the reference strip as the center, forming a point pair between the reference strip and the target strip; when the distance between the point in the target strip and the point in the reference strip is less than the distance threshold, the point pair is regarded as a valid point pair. The edge difference determination module is used to determine the quantitative index of the point cloud edge difference between the reference strip and the target strip by generating an array of differences between the points in the reference strip and the points in the target strip based on the valid point pairs.
[0014] Compared with the prior art, the present invention has the following beneficial effects: The proposed method for obtaining the edge difference of strip point clouds uses grid division to spatially discretize overlapping areas. Within each grid cell, the total number of reference strip point clouds is compared with a threshold. If the total number of reference strip point clouds is large, the data is divided into blocks. This structure and ordering of massive reference strip point cloud data provides the necessary local spatial boundaries for selecting flat areas. Only when multiple candidate points of the acquired reference strip point clouds are all located within the inscribed circle of the current grid cell and the distance range is less than the threshold, is the current grid cell considered a flat grid. This mechanism eliminates areas with large terrain undulations or rough surfaces, ensuring that subsequent calculations are based on flat areas. Simultaneously, it avoids multiple duplicate points within the same grid cell from participating in the calculation, preventing excessive local data density from causing weight bias in the overall statistical results, making the statistical results more uniform and objective. By introducing inscribed circle constraints and range threshold constraints, flat grids are effectively selected, eliminating interference from terrain undulations. Within each flat grid, the point Q1 closest to the center of the current grid cell is selected as the reference point for the flat grid. Only the most representative reference point of the flat grid is extracted for calculation, ensuring the representativeness and accuracy of the sampling point location and compressing the data volume to an extremely low level. The edge difference calculated in flat areas can more realistically reflect the systematic deviations between strips (such as elevation deviations), avoiding calculation errors caused by the unevenness of the terrain itself. Attached Figure Description
[0015] Figure 1 This is a flowchart of the method for obtaining the edge difference of strip point cloud proposed in this invention; Figure 2 This is a schematic diagram of the planar overlapping area of two point cloud stripes provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of planar overlapping region mesh division provided in an embodiment of the present invention; Figure 4 Detailed implementation steps for obtaining the edge difference of strip point cloud provided in this embodiment of the invention; Figure 5 This is a schematic diagram of a stripe with flat grid points provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the closest point association of two stripes provided in an embodiment of the present invention. Detailed Implementation
[0016] The following will refer to the appendices in the embodiments of the present invention. Figures 1-6 The technical solutions in the embodiments of the present invention will be clearly and completely described. It should be understood that the terminology used in the present invention is only for describing particular implementation methods and is not intended to limit the present invention.
[0017] Example like Figure 1 As shown, this invention proposes a method for obtaining the edge difference of strip point clouds, which includes the following steps: S1: Obtain the planar overlap region of the reference strip and the target strip, perform meshing on the planar overlap region, and determine multiple mesh cells of the planar overlap region and their corresponding mesh center points; S2: For each grid cell, when the total number of reference strip point clouds is greater than or equal to the number threshold, the reference strip point cloud is divided into data blocks; the multiple temporary candidate points closest to the grid center point of the current grid cell are obtained in each point cloud block; the multiple temporary candidate points corresponding to each point cloud block are merged, and the multiple temporary candidate points closest to the merged multiple temporary candidate points are selected as multiple candidate points of the reference strip point cloud; S3: When multiple candidate points are all located within the inscribed circle of the current grid cell and the range is less than the range threshold, the current grid cell is determined to be a flat grid; the point in the reference strip that is closest to the grid center point of the current grid cell is obtained from the flat grid. S4: Using the position of a point in the reference strip as the center, obtain the point in the target strip point cloud that is closest to the point in the reference strip, forming a point pair between the reference strip and the target strip; when the distance between a point in the target strip and a point in the reference strip is less than a distance threshold, the point pair is considered a valid point pair; S4: Based on the valid point pairs, determine the quantitative index of the point cloud edge difference between the reference strip and the target strip by generating a difference array between the points in the reference strip and the points in the target strip.
[0018] Specifically, in step S1, for a single stripe (including a reference stripe / target stripe), a grid-segment-based point cloud planar coverage acquisition method can be used to obtain the planar coverage vector of the single stripe. The obtained planar coverage vectors of the two stripes are then intersected, such as... Figure 2 As shown, the vector range of the planar overlapping region of the two stripes is obtained.
[0019] The planar overlapping region of the two strips is meshed to determine multiple mesh cells and their corresponding mesh center points, such as... Figure 3 As shown.
[0020] The maximum and minimum values of the planar overlapping area of a single strip on the X and Y axes are calculated respectively. Then, the grid row and column positions on the X and Y axes are calculated according to the grid size, and a series of evenly spaced regular grid units are divided to form a grid system covering the entire overlapping area.
[0021] The size of the mesh is determined based on the actual situation. If the size is too large, it will result in no flat mesh, and if the size is too small, it will increase the amount of computation. A balance needs to be struck between computational accuracy and efficiency.
[0022] Step S21: For each grid cell, when the total number of point clouds in the strip is greater than or equal to the threshold, the point cloud in the strip is divided into data blocks; taking each point cloud block as a unit, KD-tree can still be used to accelerate the calculation, and the distance of each point in each point cloud block from the grid center point of the current grid cell is obtained; the distances are sorted from smallest to largest, and the points ranked at the top in each point cloud block are selected. k One temporary candidate site; Within each point cloud block, sort the top-ranked blocks. k The temporary candidate points are merged, sorted in ascending order of distance, and the points with the highest distances are selected. k 1 temporary candidate point, and record the obtained points. k The distance between temporary candidate points.
[0023] Sort the merged temporary candidate points in ascending order of distance, and select the highest-ranked ones. k A temporary candidate point, as a point cloud of the strip. k There are 10 candidate points.
[0024] like Figure 4 As shown, when the total number of points in the strip point cloud is less than the threshold, the distance from each point in the strip point cloud to the center point of the current grid cell is directly obtained by building a KD tree; and the distances are sorted from smallest to largest, and the points with the highest distances are selected. k One candidate point, as a point cloud of the stripe. k There are 10 candidate points.
[0025] Step S22: When the bands are a bit of cloud k If the distance to the last candidate point in the sorted list is less than half the mesh size of the current mesh cell, then a decision is made. k All candidate points are located within the inscribed circle of the current grid cell; Determine the location within the inscribed circle of the current mesh cell. k The elevation values of each candidate point are used as the range of the current grid cell, with the difference between the maximum and minimum elevation values being the range of the current grid cell. When the range is less than the range threshold, the current grid cell is determined to be a flat grid.
[0026] The step of determining a flat grid avoids multiple duplicate points within the same grid cell from participating in the calculation, prevents excessive local data from causing weight bias in the overall statistical results, and makes the statistical results more uniform and objective.
[0027] This invention sets a threshold for the total number of point clouds, ensuring that processing is only performed when the point cloud density meets the requirements, thus avoiding unreliable calculation results due to data sparsity. The point closest to the grid center is selected as the reference point Q1, ensuring the representativeness and accuracy of the sampling point locations.
[0028] In step S3, k Among the candidate points, the point closest to the center point of the current grid cell is selected as Q1, the point closest to the center point of the current grid cell in the flat grid.
[0029] Point Q1 satisfies both the inscribed circle constraint and the range threshold constraint. Similarly, a search is performed on each of the divided grid cells, such as... Figure 5 As shown, obtain flat grids that simultaneously meet the above two filtering conditions, and select the point of the flat grid that is closest to the grid center point of the current grid cell, and form an array Q together with point Q1.
[0030] This invention completely abandons the existing technology's reliance on a "registration first, comparison later" process. By introducing inscribed circle constraints and range threshold constraints, it effectively filters out flat grids and eliminates interference from terrain undulations. Only when all candidate points are within the inscribed circle of the grid and the distance range is less than the threshold is the grid considered flat. This mechanism eliminates areas with large terrain undulations or rough surfaces, ensuring that subsequent calculations are based on flat areas. Calculating the edge difference in flat areas more realistically reflects the systematic deviations between strips (such as elevation deviations), avoiding calculation errors caused by the unevenness of the terrain itself.
[0031] In step S4, with the position of point Q1 as the center, the point Q2 closest to point Q1 in the point cloud of strip two (target strip) is obtained by KD search; When the total number of two-point clouds in the stripe is greater than or equal to the threshold, the two-point clouds in the stripe are divided into data blocks. For each point cloud block, a KD-tree is built to obtain the distance to point Q1 in each block. The distances are sorted from smallest to largest, and the top-ranked points in each block are selected. k A temporary candidate point; sort the top-ranked points in each point cloud block. k The candidate points are merged and sorted in ascending order of distance. The first-ranked point is taken as the closest point Q2 to point Q1 in the two-point cloud of the strip.
[0032] When the total number of point clouds in the two stripes is less than the number threshold, the point Q2 closest to point Q1 in the two stripes is directly obtained through KD search, forming a point pair between strip one and strip two; When the distance between point Q2 and point Q1 is less than the distance threshold, the point pair is considered a valid point pair.
[0033] Similarly, a search is performed on each of the divided grid cells, such as... Figure 6 As shown, the flat grids that satisfy the inscribed circle constraint and the difference threshold constraint in each grid cell of strip one are obtained respectively, and the point Q1 that is closest to the grid center point of the current grid cell is obtained; in the point cloud of strip two, the point Q2 that satisfies point Q1 is selected, and points Q1 and Q2 are combined to form an array Q.
[0034] By filtering valid point pairs using a distance threshold, outliers that were too far apart were eliminated, further ensuring the quality of the data.
[0035] This invention employs a regular grid to spatially discretize overlapping regions, extracting only the most representative flat grid within each grid cell for computation, thus compressing the data volume to an extremely low level. This method overcomes the technical shortcomings of traditional point-by-point comparison, which requires processing millions or even tens of millions of point pairs within overlapping areas, resulting in enormous computational and storage pressure. Furthermore, this method reduces the computational complexity of nearest neighbor queries from linear to logarithmic levels through KD-tree spatial indexing and a block-based computation strategy. Practical tests show that this method can rapidly assess the elevation differences of billions of point cloud strips on ordinary workstations, solving the technical bottleneck of computationally incapable of handling massive amounts of data.
[0036] In step S4, based on the formed valid point pairs, the quantitative index of the point cloud edge connection difference between strip one and strip two is determined by generating a difference array between Q1 and Q2, specifically including: Based on the formed valid point pairs, generate the difference array between Q1 and Q2. D for: Use the quantile method to remove the difference array D Gross differences in: 1) Calculation D The first quartile (q1): the boundary value of the first 25% of the data after the data is arranged in ascending order; 2) Calculation D The third quartile (q3): the boundary value of the first 75% of the data after the data is arranged in ascending order; 3) Calculate the interquartile range (iqr): ; 4) Determine a reasonable data range: usually take ; 5) Filter out data within a reasonable range and remove outlier values that exceed the outlier threshold. Finally, the mean and standard error of the statistical difference array are used as quantitative indicators of the edge difference between strip one and strip two.
[0037] By generating a difference array between point Q1 and point Q2, the abstract point cloud overlap quality is transformed into a specific quantitative indicator, providing intuitive and operable data support for point cloud quality assessment, stitching accuracy inspection, and subsequent correction processing.
[0038] This invention also proposes a strip point cloud edge difference acquisition system, comprising: The mesh generation module is used to obtain the planar overlapping area of the reference strip and the target strip, perform mesh generation on the planar overlapping area, and determine multiple mesh cells and their corresponding mesh center points in the planar overlapping area. The flat grid generation module is used to divide the reference strip point cloud into blocks for each grid cell when the total number of reference strip point clouds is greater than or equal to the number threshold; obtain the multiple temporary candidate points that are closest to the grid center point of the current grid cell in each point cloud block; merge the multiple temporary candidate points corresponding to each point cloud block, and select the multiple temporary candidate points that are closest to each other from the merged multiple temporary candidate points as multiple candidate points of the reference strip point cloud; The valid point pair generation module is used to determine that the current grid cell is a flat grid when multiple candidate points are all located within the inscribed circle of the current grid cell and the range is less than the range threshold; to obtain the point in the reference strip that is closest to the grid center point of the current grid cell; to obtain the point in the target strip point cloud that is closest to the point in the reference strip, using the position of the point in the reference strip as the center, forming a point pair between the reference strip and the target strip; when the distance between the point in the target strip and the point in the reference strip is less than the distance threshold, the point pair is regarded as a valid point pair. The edge difference determination module is used to determine the quantitative index of the point cloud edge difference between the reference strip and the target strip by generating an array of differences between the points in the reference strip and the points in the target strip based on the valid point pairs.
[0039] This invention requires only localized micro-scale flatness of the ground surface, without any corners, edges, or artificial markers. This makes the method seamlessly applicable to all terrain scenarios, from urban built-up areas to uninhabited areas, and from mountains to plains, greatly expanding the engineering applicability of strip edge difference assessment.
[0040] The method proposed in this invention effectively avoids the influence of complex terrain and noise on the results through a specific flat area screening strategy and multiple threshold constraints. It solves the problem that the calculation of edge difference is greatly affected by terrain undulation in traditional methods, thereby significantly improving the accuracy and robustness of obtaining the edge difference between airborne radar point cloud strips.
[0041] Therefore, the method proposed in this invention achieves high-precision extraction, high-efficiency calculation, global adaptability and high robust output of the edge difference of strip point cloud without relying on registration, manual targets or significant geometric features, thus overcoming the core bottlenecks of existing technologies in practical engineering applications.
[0042] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.
[0043] Furthermore, unless otherwise stated, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. All references to this specification are incorporated by way of citation to disclose and describe methods relating to those references. In the event of any conflict with any incorporated reference, the content of this specification shall prevail.
Claims
1. A method for obtaining the edge difference of a strip point cloud, characterized in that, Includes the following steps: Obtain the planar overlap region of the reference strip and the target strip, perform meshing on the planar overlap region, and determine multiple mesh cells and their corresponding mesh center points in the planar overlap region; For each grid cell, when the total number of reference strip point clouds is greater than or equal to the number threshold, the reference strip point cloud is divided into data blocks. Obtain multiple temporary candidate points in each point cloud block that are closest to the center point of the current grid cell; Multiple temporary candidate points corresponding to each point cloud block are merged, and the closest temporary candidate points are selected from the merged temporary candidate points as multiple candidate points for the reference strip point cloud. When multiple candidate points are all located within the inscribed circle of the current grid cell and the range is less than the range threshold, the current grid cell is determined to be a flat grid. Obtain the point in the reference strip that is closest to the center point of the current grid cell in the flat grid; Using the position of a point in the reference strip as the center, obtain the point in the target strip point cloud that is closest to the point in the reference strip, forming a point pair between the reference strip and the target strip; when the distance between a point in the target strip and a point in the reference strip is less than a distance threshold, the point pair is considered a valid point pair; Based on the valid point pairs, a quantitative index of the point cloud edge difference between the reference strip and the target strip is determined by generating a difference array between the points in the reference strip and the points in the target strip.
2. The method for obtaining the edge difference of strip point cloud according to claim 1, characterized in that, For each grid cell, when the total number of reference strip point clouds is greater than or equal to the number threshold, the reference strip point cloud is divided into data blocks; and multiple temporary candidate points that are closest to the grid center point of the current grid cell are obtained in each point cloud block. Multiple temporary candidate points corresponding to each point cloud block are merged, and the closest temporary candidate points are selected from the merged temporary candidate points as multiple candidate points for the reference strip point cloud. Specifically, this includes: For each grid cell, when the total number of reference strip point clouds is greater than or equal to the number threshold, the reference strip point cloud is divided into data blocks to obtain multiple point cloud blocks. Using each point cloud block as a unit, the distance from each point in the current point cloud block to the center point of the current grid cell is obtained by building a KD tree search; then the distances are sorted from smallest to largest, and the points with the highest distances are selected. k A temporary candidate point is used as the corresponding point cloud block. k One temporary candidate site; Correspondingly obtain the other point cloud blocks k One temporary candidate site; The corresponding points in each point cloud block k The temporary candidate points are merged to obtain multiple temporary candidate points after merging. Sort the merged temporary candidate points in ascending order of distance, and select the highest-ranked ones. k A temporary candidate point is used as a reference strip point cloud. k There are 10 candidate points.
3. The method for obtaining the edge difference of strip point cloud according to claim 1, characterized in that, When all candidate points are located within the inscribed circle of the current grid cell and the range is less than the range threshold, the current grid cell is determined to be a flat grid. Obtain the point in the reference strip that is closest to the center point of the current grid cell in the flat grid, specifically including: When the distance to the last candidate point in the sorting of multiple candidate points in the reference strip point cloud is less than half the grid size of the current grid cell, it is determined that all multiple candidate points are located within the inscribed circle of the current grid cell. Determine the elevation values of multiple candidate points located within the inscribed circle of the current grid cell, and use the difference between the maximum and minimum elevation values as the range of the current grid cell; When the range is less than the range threshold, the current grid cell is determined to be a flat grid. The point in the reference strip that is closest to the center point of the current grid cell in the flat grid is obtained by KD tree search.
4. The method for obtaining the edge difference of strip point cloud according to claim 1, characterized in that, The step of obtaining the point in the target strip that is closest to the point in the reference strip from the point cloud of the target strip, using the position of the point in the reference strip as the center, to form a point pair between the reference strip and the target strip, specifically includes: Using the planar position of a point in the reference strip as the center, a KD search is used to obtain the point in the target strip that is closest to the point in the reference strip in the target strip point cloud, forming a point pair between the reference strip and the target strip.
5. The method for obtaining the edge difference of strip point cloud according to claim 1, characterized in that, The step of determining the quantitative index of the point cloud edge difference between the reference strip and the target strip by generating a difference array between points in the reference strip and points in the target strip based on valid point pairs specifically includes: Based on the valid point pairs, the difference array D between the points in the reference strip and the points in the target strip is generated as follows: in, Q 1 represents a point in the reference strip. Q 2 represents a point within the target strip; The quantile method is used to remove outliers from the difference array D, and the mean and mean error of the difference array are calculated as quantitative indicators of the point cloud edge difference between the reference strip and the target strip.
6. The method for obtaining the edge difference of strip point cloud according to claim 2, characterized in that, For each grid cell, the method further includes: When the total number of reference strip point clouds is less than the number threshold, a KD-tree search is used to obtain the distance of each point in the reference strip point cloud from the center point of the current grid cell; the distances are sorted from smallest to largest, and the points with the highest distances are selected. k 10 candidate points, used as reference strip point clouds k There are 10 candidate points.
7. The method for obtaining the edge difference of strip point cloud according to claim 1, characterized in that, The process of obtaining the planar overlap region of the reference strip and the target strip, dividing the planar overlap region into a grid, and determining multiple grid cells and their corresponding grid center points for the planar overlap region specifically includes: The planar coverage vectors of the reference strip and the target strip are obtained using a point cloud planar coverage method based on grid segmentation. The intersection operation is performed on the planar coverage vectors of the obtained reference strip and target strip to obtain the planar overlapping area vector range of the reference strip and target strip; Based on the vector range of the planar overlapping region, the planar overlapping region of the reference strip and the target strip is meshed to determine multiple mesh cells corresponding to the planar overlapping region and their corresponding mesh center points.
8. A system for acquiring edge difference of strip point cloud, characterized in that, include: The mesh generation module is used to obtain the planar overlapping area of the reference strip and the target strip, perform mesh generation on the planar overlapping area, and determine multiple mesh cells and their corresponding mesh center points in the planar overlapping area. The flat grid generation module is used to divide the reference strip point cloud into data blocks for each grid cell when the total number of reference strip point clouds is greater than or equal to the number threshold. Obtain multiple temporary candidate points in each point cloud block that are closest to the center point of the current grid cell; Multiple temporary candidate points corresponding to each point cloud block are merged, and the closest temporary candidate points are selected from the merged temporary candidate points as multiple candidate points for the reference strip point cloud. The effective point pair generation module is used to determine that the current grid cell is a flat grid when multiple candidate points are all located inside the inscribed circle of the current grid cell and the range is less than the range threshold. Obtain the point in the reference strip that is closest to the center point of the current grid cell in the flat grid; Using the position of a point in the reference strip as the center, obtain the point in the target strip point cloud that is closest to the point in the reference strip, forming a point pair between the reference strip and the target strip; when the distance between a point in the target strip and a point in the reference strip is less than a distance threshold, the point pair is considered a valid point pair; The edge difference determination module is used to determine the quantitative index of the point cloud edge difference between the reference strip and the target strip by generating an array of differences between the points in the reference strip and the points in the target strip based on the valid point pairs.