Pipeline laser radar point cloud processing method based on dynamic block least square method
By using the dynamic block least squares method, the noise and unevenness issues of pipeline point cloud data were resolved, generating high-quality 3D point cloud data and ensuring the geometric fidelity and integrity of the pipeline 3D model.
Patent Information
- Application Number
- CN202511743154.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-25
- Publication Date
- 2026-02-17
AI Technical Summary
The original pipeline point cloud data has problems with noise and uneven distribution, which leads to defects in the reconstructed 3D pipeline model and makes it difficult to obtain high-quality pipeline point cloud data.
A processing method based on dynamic block least squares is adopted, including bilateral filtering for noise reduction, sector segmentation, dynamic adjustment of sampling range and moving least squares fitting, resampling and axial densification, to generate high-quality 3D point cloud data.
It effectively removed noise, homogenized the point cloud distribution, and solved the problem of uneven density of pipe point clouds in the radial and axial directions, generating a 3D pipe model with clear texture and high geometric fidelity.
Smart Images
Figure CN121544489A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of point cloud data processing technology, and more specifically, relates to a pipeline lidar point cloud processing method based on dynamic block least squares method. Background Technology
[0002] For urban drainage pipelines, point cloud data can be used to reconstruct 3D models of the pipelines. However, raw pipeline point cloud data suffers from noise and severely uneven distribution (especially with large differences in density between the axial and radial directions). Directly using raw pipeline point cloud data for reconstruction leads to numerous defects in the reconstructed 3D pipeline model. High-quality pipeline point cloud data can ensure the geometric fidelity and integrity of the 3D pipeline model. How to obtain high-quality pipeline point cloud data is a pressing technical problem that needs to be solved in this field. Summary of the Invention
[0003] In view of the shortcomings of the existing technology, the purpose of this application is to achieve the acquisition of high-quality pipeline point cloud data.
[0004] To achieve the above objectives, in a first aspect, this application provides a pipeline lidar point cloud processing method based on dynamic block least squares, the method comprising: Bilateral filtering denoising is performed on the original two-dimensional lidar point cloud data of the pipe cross-section to obtain the denoised two-dimensional point cloud data of the cross-section. Based on the denoised two-dimensional point cloud data of the cross section, the center of the point cloud is determined and the point cloud is divided into multiple sector blocks; The sampling range for each sector is determined by using a dynamic block method. The dynamic block method uses the size of the sector as a unit and dynamically expands from the center line of the sector to both sides to determine the sampling range. The point cloud within the sampling range corresponding to each sector block is resampled using the least squares method to obtain the resampled points of each sector block and generate the resampled two-dimensional point cloud data of the cross section. Two-dimensional point cloud data of the pipeline is synthesized by upsampling along the pipeline axis after resampling of multiple cross-sections of the pipeline.
[0005] In one possible implementation, the above-mentioned bilateral filtering denoising process is performed on the original two-dimensional lidar point cloud data of the pipe cross-section to obtain the denoised two-dimensional point cloud data of the cross-section, including updating each point in the original two-dimensional lidar point cloud data of the cross-section using the following formula. The position is obtained as a new point. : ; ; in, For point-based bilateral filtering An estimated unit undirected normal vector, Point The neighborhood, for points within, Point With point Geometric distance between them Representing vectors In the normal vector Projection on; and It is a weighting function.
[0006] In one possible implementation, the aforementioned denoised two-dimensional point cloud data based on cross-sections is used to determine the point cloud center and divide the point cloud into multiple sector blocks, including: Based on the denoised two-dimensional point cloud data of the cross section, the center position of the two-dimensional point cloud is determined by minimizing the error between the point set and the fitted circle; Based on the center position of the two-dimensional point cloud, the denoised two-dimensional point cloud data is divided into multiple sector blocks with equal angles.
[0007] In one possible implementation, the above-mentioned method of determining the sampling range corresponding to each sector block through dynamic block generation includes: Set an axis with the center of the 2D point cloud as the origin as the pointer, and point the pointer to the center line of the sector. Select the area around the pointer from... An initial region consisting of several sector-shaped blocks serves as the initial turntable, with the pointer as the axis of symmetry. It is a positive integer; Using the pointer as the axis of symmetry, the turntable is divided into a left disk and a right disk. The range of the left and right disks is dynamically expanded in units of the size of the sector block until the point count constraint is met. At the same time, the angle between the pointer and any edge of the turntable is not more than 90 degrees. The turntable that meets the constraint is used as the sampling range corresponding to the sector block.
[0008] In one possible implementation, the point constraint is determined by the following formula: ; in, Indicates left disk The number of points inside, Indicates right-hand drive The number of points within; and The threshold number of points; This is the threshold for the difference in the number of points.
[0009] In one possible implementation, the point cloud within the sampling range corresponding to each sector block is resampled using the least squares method to obtain the resampled points of each sector block and generate resampled two-dimensional point cloud data of the cross section. This includes obtaining the resampled points of the sector blocks through the following steps: Based on the point cloud data within the sampling range corresponding to the sector block, the moving least squares method is used to perform polynomial curve fitting on the point cloud to obtain the fitted curve. Based on the intersection of the fitted curve and the vertical axis of the pointer coordinate system, the resampling points corresponding to the sector blocks are determined; The pointer coordinate system is a rectangular coordinate system with the pointer as the vertical axis.
[0010] In one possible implementation, the loss function used for polynomial curve fitting is as follows: ; in, Let be the coefficient vector of the polynomial to be solved. This indicates the number of the sampling point within the sampling range corresponding to the sector block. This represents the total number of sampling points within the sampling range corresponding to the sector block; This represents the coordinates of the sampling point in the pointer coordinate system. The x-axis coordinates of the sampling points. These are the vertical axis coordinates of the sampling points. Represents the fitted polynomial curve. This is the weighting function.
[0011] In one possible implementation, The formula is as follows: ; in, This represents the length of the arc segment from the sampling point to the pointer. This represents the maximum distance between adjacent points within the turntable.
[0012] Secondly, this application provides a pipeline lidar point cloud processing device based on dynamic block least squares method, comprising: The denoising module is used to perform bilateral filtering denoising on the original two-dimensional lidar point cloud data of the cross-section of the pipe to obtain the denoised two-dimensional point cloud data of the cross-section. The block partitioning module is used to determine the center of the point cloud and divide the point cloud into multiple sector blocks based on the denoised two-dimensional point cloud data of the cross section. The sampling range determination module is used to determine the sampling range corresponding to each sector block through a dynamic block method. The dynamic block method uses the size of the sector block as the unit and dynamically expands from the center line of the sector block to both sides to determine the sampling range. The resampling module is used to resample the point cloud within the sampling range corresponding to each sector block using the least squares method, obtain the resampled points of each sector block, and generate the resampled two-dimensional point cloud data of the cross section. The 3D point cloud data synthesis module is used to synthesize 3D point cloud data of the pipeline by upsampling the 2D point cloud data after resampling multiple cross-sections of the pipeline along the pipeline axis.
[0013] Thirdly, this application provides an electronic device, including: a memory and one or more processors; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code including computer instructions; the one or more processors invoke the computer instructions to cause the electronic device to perform the method described in the first aspect or any possible implementation of the first aspect.
[0014] Overall, the technical solutions conceived in this application have the following beneficial effects compared with the prior art: By employing a bilateral filtering algorithm, noise can be effectively removed while better preserving key details such as the geometry and edges of the pipe surface. Through sector segmentation, dynamic adjustment of the sampling range, and moving least squares fitting, the problem of severe density unevenness in the radial and axial directions of the LiDAR point cloud in the pipe can be effectively solved. The algorithm adaptively handles the local sparsity of the point cloud, restoring the structure of sparse regions while preserving details in dense areas, resulting in a more uniform point cloud distribution and avoiding reconstruction defects caused by local sparsity. Therefore, through the processing flow of "denoising-segmentation-dynamic resampling-axial densification," the problems of noise and severe uneven distribution (especially the large density difference between axial and radial directions) in the original pipe point cloud are systematically solved, enabling the output of high-quality 3D point cloud data. This provides a foundation for generating a clear-textured, geometrically accurate 3D pipe model. Attached Figure Description
[0015] Figure 1 This is a flowchart illustrating the pipeline lidar point cloud processing method based on dynamic block least squares provided in the embodiments of this application. Figure 2 This is a diagram illustrating the bidirectional filtering and denoising effect provided in an embodiment of this application. Figure 3 This is a schematic diagram of the initialization of the sampling range provided in an embodiment of this application; Figure 4This is a schematic diagram of moving least squares resampling provided in an embodiment of this application; Figure 5 This is a schematic diagram of a uniformly distributed three-dimensional point cloud dataset provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of the pipeline lidar point cloud processing device based on the dynamic block least squares method provided in the embodiments of this application; Figure 7 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation
[0016] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0017] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.
[0018] In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more, for example, multiple processing units means two or more processing units, multiple elements means two or more elements, etc.
[0019] The embodiments of this application are described below with reference to the accompanying drawings.
[0020] Figure 1 This is a flowchart illustrating the pipeline lidar point cloud processing method based on dynamic block least squares provided in this application embodiment, as shown below. Figure 1 As shown, the method includes the following steps S1 to S5.
[0021] Step S1: Perform bilateral filtering denoising on the original two-dimensional lidar point cloud data of the pipe cross-section to obtain the denoised two-dimensional point cloud data of the cross-section. Step S2: Based on the denoised two-dimensional point cloud data of the cross section, determine the center of the point cloud and divide the point cloud into multiple sector blocks; Step S3: Determine the sampling range corresponding to each sector block using a dynamic block method. The dynamic block method uses the size of the sector block as the unit and dynamically expands from the center line of the sector block to both sides to determine the sampling range. Step S4: Use the least squares method to resample the point cloud within the sampling range corresponding to each sector block, obtain the resampled points of each sector block, and generate the resampled two-dimensional point cloud data of the cross section (uniformed two-dimensional point cloud). Step S5: Based on the resampled two-dimensional point cloud data of multiple cross-sections of the pipeline, upsample along the pipeline axis to synthesize three-dimensional point cloud data of the pipeline.
[0022] Understandably, by employing a bilateral filtering algorithm, noise can be effectively removed while better preserving key details such as the geometry and edges of the pipe surface. By using sector segmentation, dynamically adjusting the sampling range, and moving least squares fitting, the problem of severe density unevenness of LiDAR point clouds in the radial and axial directions of the pipe can be effectively solved. The local sparsity of the point cloud can be adaptively processed, preserving details in dense areas while restoring the structure of sparse areas, resulting in a more uniform point cloud distribution and avoiding defects in reconstruction caused by local sparsity.
[0023] In one possible implementation, step S1 specifically includes S11 to S13.
[0024] S11. The raw point cloud data was acquired by a LiDAR and meter counter mounted on a pipe-crawling robot. The LiDAR moved along the pipe axis, acquiring the two-dimensional cross-sectional point cloud inside the pipe in real time. A complete ring of two-dimensional point cloud data was generated every 5 cm (as raw two-dimensional LiDAR point cloud data of a cross-section). Each ring of point cloud reflects the spatial structure of the pipe cross-section at that location. Multiple consecutive rings of point cloud are arranged along the pipe axis to form a two-dimensional point cloud sequence with spatial order, providing basic data for subsequent denoising and resampling processing.
[0025] S12, bilateral filtering is applied to each point. Estimate a unit undirected normal vector The normal vector passes through the point. neighborhood The calculation yielded: ; In the formula, This is a point cloud dataset constructed based on raw two-dimensional lidar point cloud data from a single cross-section. is the neighborhood radius.
[0026] S13, Update Point Get a new position right Each point in After performing the position update, the denoised 2D point cloud data of the cross-section is obtained: ; Point position update It is calculated by the following formula: ; In the formula, For point neighborhood points within; For point With point The geometric distance between them; For vectors In the normal vector Projection on; and Based on spatial distance variance and normal offset variance The central Gaussian weighting function with parameters is used to weight the data based on geometric distance and normal offset.
[0027] The effect of bilateral filtering on the processing of the original point cloud is as follows: Figure 2 As shown.
[0028] In one possible implementation, step S2 specifically includes S21 to S22.
[0029] S21. Determine the center position of the two-dimensional point cloud by minimizing the error between the point set and the fitted circle. and radius The fitting loss function used is: ; In the formula, The point set (the point set determined from the denoised 2D point cloud data based on the cross-section) is the th The coordinates of the points To fit the radius of the circle, This represents the total number of points in the point set.
[0030] S22, Sector Division: Using the center point obtained from the fitting in S21... Centered on the denoised two-dimensional point cloud data, the data is divided into: A sector-shaped area with equal angles, in which An adjustable parameter for controlling resampling density.
[0031] Using the center of the fitted circle as the center, the point cloud data is divided into... There are several parts. Among them, It is an adjustable parameter that can be adjusted. The value allows for flexible control of the point cloud resampling density. A higher value... A higher value allows for more detailed resampling, suitable for high-precision reconstruction, while a lower value... Values help improve computational efficiency and reduce redundant calculations.
[0032] The angular range of each sector is: The centerline of each block is a ray running from the center of the circle to the edge of the block, and this ray lies on the axis of symmetry of the sector block. In this way, the point cloud data is divided into multiple sector blocks, providing a basis for subsequent dynamic resampling.
[0033] In one possible implementation, step S3 specifically includes performing the following operations for each sector block (steps S31 to S32) to obtain the sampling range corresponding to each sector block.
[0034] S31. Initialize a line centered on the point cloud. The axis with the origin is used as a "pointer," and the pointer points to the center line of the sector, with the selection around the pointer being... An initial region consisting of several sector-shaped blocks serves as the initial "wheel" (or dynamic block), in which... This is an adjustable parameter.
[0035] S32, Sampling range initialization as follows Figure 3 As shown, the turntable is divided into a left and right disk with the pointer as the axis of symmetry. (For the initial turntable, the left turntable) Include (a sector) and the right disk (For the initial turntable, the right turntable) Include (Each sector-shaped area); using the size of the sector-shaped area as the unit of dynamic expansion, the range of the left and right disks is dynamically expanded until the following point constraint conditions are met, while ensuring the angle between the pointer and any edge of the disk is maintained. (The angle between the pointer and the left edge of the dial) and The angle between the pointer and the right edge of the turntable should not exceed 90 degrees to obtain a turntable that satisfies the constraints. ; In the formula, and The number of dots in the left and right disks; and The threshold number of points; This is the threshold for the difference in the number of points.
[0036] To facilitate subsequent polynomial curve fitting, the point cloud data within the turntable can be transformed from the LiDAR coordinate system to a pointer coordinate system with the pointer as the Y-axis (or vertical axis). The LiDAR point cloud data is mapped to the point cloud dataset in the pointer coordinate system through rotation and translation operations. The datasets on both coordinate axes can be represented as follows: ; ; In the formula, This is a point cloud dataset in pointer coordinate system. The point cloud dataset within the turntable; and These are the coordinates in the pointer coordinate system and the coordinates in the lidar coordinate system, respectively.
[0037] The mapping relationship can then be represented as: ; ; In the formula, The angle between the pointer and the center line of the turntable.
[0038] In one possible implementation, step S4 is as follows: Figure 4 As shown, the specific steps include performing the following resampling operations (steps S41 to S42) on the sampling range corresponding to each sector block to obtain the resampling points corresponding to each sector block, and then generating resampled two-dimensional point cloud data of the cross section based on the resampling points corresponding to each sector block.
[0039] S41. Based on the point cloud data within the sampling range corresponding to the sector block, the moving least squares (MLS) method is used to perform polynomial curve fitting on the point cloud, and its loss function is defined as: ; In the formula, The fitted polynomial curve, Let be the coefficient vector of the polynomial to be solved; The coordinates of the point in the pointer coordinate system; Let be the weighting function, which is defined as: ; Among them, distance The length of the arc segment from the sampling point to the pointer is calculated by the following formula: ; In the formula, For point The angle between the pointer and the direction of the pointer; Center of the point cloud; parameters The value is dynamically defined as the maximum distance between all adjacent points within the current turntable.
[0040] S42. Select the intersection point of the fitted curve in S41 and the Y-axis of the pointer coordinate system, i.e. This serves as a new resampling point at the pointer position (i.e., the resampling point corresponding to the sector block), where The constant term is used to fit the polynomial.
[0041] It is understandable that after performing steps S41 to S42 for the sampling range corresponding to each sector block, a preset number can be generated. With uniformly distributed two-dimensional resampling points, based on these n resampling points, resampled two-dimensional point cloud data of the cross section can be generated.
[0042] In one possible implementation, step S5 specifically includes: By resampling two-dimensional point cloud data from multiple cross-sections of the pipe, sorting it along the pipe axis, and performing linear interpolation (to increase the point cloud density along the axis), a uniformly distributed three-dimensional point cloud dataset can be generated, such as... Figure 5 As shown.
[0043] In summary, the processing flow of "denoising-blocking-dynamic resampling-axial densification" systematically solves the problems of noise and severe uneven distribution (especially the large difference in density between the axial and radial directions) in the original pipeline point cloud, and can output high-quality 3D point cloud data, providing a foundation for generating pipeline 3D models with clear texture and high geometric fidelity.
[0044] The following describes the pipeline lidar point cloud processing device based on dynamic block least squares method provided in this application. The pipeline lidar point cloud processing device based on dynamic block least squares method described below can be referred to in correspondence with the pipeline lidar point cloud processing method based on dynamic block least squares method described above.
[0045] Figure 6 This is a schematic diagram of the structure of the pipeline lidar point cloud processing device based on the dynamic block least squares method provided in the embodiments of this application, as shown below. Figure 6 As shown, the device includes: a denoising module 10, a block division module 20, a sampling range determination module 30, a resampling module 40, and a 3D point cloud data synthesis module 50. Wherein: The denoising module 10 is used to perform bilateral filtering denoising on the original two-dimensional lidar point cloud data of the cross-section of the pipe to obtain the denoised two-dimensional point cloud data of the cross-section. Block partitioning module 20 is used to determine the center of the point cloud and divide the point cloud into multiple sector blocks based on the denoised two-dimensional point cloud data of the cross section. The sampling range determination module 30 is used to determine the sampling range corresponding to each sector block through a dynamic block method. The dynamic block method is based on the size of the sector block and dynamically expands from the center line of the sector block to both sides to determine the sampling range. The resampling module 40 is used to resample the point cloud within the sampling range corresponding to each sector block using the least squares method, obtain the resampling points of each sector block and generate the resampled two-dimensional point cloud data of the cross section. The 3D point cloud data synthesis module 50 is used to synthesize 3D point cloud data of the pipeline by upsampling the 2D point cloud data after resampling multiple cross-sections of the pipeline along the pipeline axis.
[0046] It is understood that the detailed functional implementation of each of the above units / modules can be found in the description in the aforementioned method embodiments, and will not be repeated here.
[0047] It should be understood that the above-described device is used to execute the methods in the above embodiments. The implementation principle and technical effect of the corresponding program modules in the device are similar to those described in the above methods. The working process of the device can be referred to the corresponding process in the above methods, and will not be repeated here.
[0048] Based on the methods in the above embodiments, this application provides an electronic device, such as... Figure 7 As shown, the electronic device may include a processor 810, a communications interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communications interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute the methods in the above embodiments.
[0049] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.
[0050] Based on the methods in the above embodiments, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to execute the methods in the above embodiments.
[0051] Based on the methods in the above embodiments, this application provides a computer program product that, when run on a processor, causes the processor to execute the methods in the above embodiments.
[0052] It is understood that the processor in the embodiments of this application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor can be a microprocessor or any conventional processor.
[0053] The method steps in this application embodiment can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an ASIC.
[0054] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).
[0055] It is understood that the various numerical designations used in the embodiments of this application are merely for the convenience of description and are not intended to limit the scope of the embodiments of this application.
[0056] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.
Claims
1. A method for processing point clouds of a tube lidar based on dynamic block least squares, characterized in that, The method comprises the following steps: Bilateral filtering denoising is performed on the original two-dimensional laser radar point cloud data of the cross section of the pipeline to obtain denoised two-dimensional point cloud data of the cross section; Based on the denoised two-dimensional point cloud data of the cross section, the center of the point cloud is determined and the point cloud is divided into a plurality of sector blocks; The sampling range corresponding to each sector block is determined by a dynamic block method, which is a method of dynamically expanding from the center line of the sector block to both sides in units of the size of the sector block to determine the sampling range; The point cloud in the sampling range corresponding to each sector block is resampled by the least square method to obtain the resampled points of each sector block and generate the resampled two-dimensional point cloud data of the cross section; Based on the resampled two-dimensional point cloud data of the plurality of cross sections of the pipeline, the pipeline three-dimensional point cloud data is synthesized by upsampling along the pipeline axial direction.
2. The method of claim 1, wherein, The original two-dimensional laser radar point cloud data of the cross section of the pipeline is subjected to bilateral filtering denoising processing to obtain denoised two-dimensional point cloud data of the cross section, including updating the position of each point in the original two-dimensional laser radar point cloud data of the cross section through the following formula to obtain a new point : ； ； wherein is a bilateral filter for a point estimated from a unit non-directional vector, denotes a neighborhood of a point , is a point within the neighborhood, denotes a geometric distance between a point and a point , denotes a projection of a vector onto a normal vector ; and are weight functions.
3. The method of claim 1, wherein, The method comprises the following steps: Based on the denoised two-dimensional point cloud data of the cross section, the center of the point cloud is determined and the point cloud is divided into a plurality of sector blocks, which comprises the following steps: The center position of the two-dimensional point cloud is determined by minimizing the error between the point set and the fitted circle based on the denoised two-dimensional point cloud data of the cross section; 4. The method of claim 1, wherein, Based on the center position of the two-dimensional point cloud, the denoised two-dimensional point cloud data is divided into a plurality of sector blocks with equal angles. An axis with the center position of the two-dimensional point cloud as the origin is set as a pointer, and the pointer points to the center line of the sector block, and an initial area composed of a plurality of sector blocks is selected around the pointer as an initial carousel, and the initial carousel takes the pointer as the axis of symmetry, is a positive integer; The sampling range corresponding to each sector block is determined by a dynamic block method, which comprises the following steps:
5. The method of claim 4, wherein, The range of the left and right discs is dynamically expanded in units of the size of the sector block until it meets the point number constraint condition, while ensuring that the angle between the pointer and any edge of the disc does not exceed 90 degrees, to obtain the disc that meets the constraint as the sampling range corresponding to the sector block. ； wherein, represents the number of points in the left disc represents the number of points in the right disc represents the number of points in the right disc represents the number of points in the right disc and is a point number threshold value; is a point number difference threshold value.
6. The method of claim 4, wherein, The point number constraint condition is determined by the following formula: The resampled points of the sector block are obtained by the following steps: Based on the point cloud data in the sampling range corresponding to the sector block, a polynomial curve fitting is performed on the point cloud by using the moving least square method to obtain a fitting curve; Based on the intersection of the fitting curve and the vertical axis of the pointer coordinate system, the resampled points corresponding to the sector block are determined; 7. The method of claim 6, wherein, The pointer coordinate system is a rectangular coordinate system with the pointer as the vertical axis. ； wherein, is a polynomial coefficient vector to be solved, denotes the number of sampling points in the sampling range corresponding to the sector block, denotes the total number of sampling points in the sampling range corresponding to the sector block; denotes the coordinate of the sampling point in the pointer coordinate system, is the horizontal axis coordinate value of the sampling point, is the vertical axis coordinate value of the sampling point, denotes the fitted polynomial curve, is a weight function.
8. The method of claim 7, wherein, The formula is as follows: ； wherein, denotes the arc segment length of the sampling point to the pointer, denotes the maximum value of the distance between adjacent points in the turntable.
9. A pipeline lidar point cloud processing apparatus based on dynamic block least squares, characterized by, The loss function used for polynomial curve fitting is as follows: The method comprises the following steps: A denoising module is configured to perform bilateral filtering denoising on the original two-dimensional laser radar point cloud data of the cross section of the pipeline to obtain denoised two-dimensional point cloud data of the cross section; A block division module is configured to determine the center of the point cloud and divide the point cloud into a plurality of sector blocks based on the denoised two-dimensional point cloud data of the cross section; A sampling range determination module is configured to determine the sampling range corresponding to each sector block by a dynamic block method, which is a method of dynamically expanding from the center line of the sector block to both sides in units of the size of the sector block to determine the sampling range; The resampling module is configured to resample the point cloud in the sampling range corresponding to each sector block by using a least square method, to obtain resampled points of each sector block, and to generate resampled two-dimensional point cloud data of the cross section; The three-dimensional point cloud data synthesis module is configured to synthesize three-dimensional point cloud data of the pipeline based on the resampled two-dimensional point cloud data of the plurality of cross sections of the pipeline and by upsampling along the axial direction of the pipeline.
10. An electronic device, comprising: The electronic device comprises: a memory and one or more processors; the memory is coupled to the one or more processors, and the memory is configured to store computer program codes, the computer program codes comprising computer instructions; the one or more processors invoke the computer instructions to enable the electronic device to perform the method according to any one of claims 1-8.