Point cloud data processing and refined calculation method, device and readable storage medium thereof

Abstract

The present invention proposes a method, device and readable storage medium for processing and refining point cloud data. In view of the fact that the existing technology generates point clouds with poor accuracy, insufficient refinement and many noise points, the present invention proposes an innovative solution. It uses a unique noise removal step to accurately remove noise points by using the expansion grid and custom feature growth judgment; through the point cloud segmentation and recognition step, accurate segmentation is achieved based on calculations such as local normal vectors; linear fitting, segmentation merging and resampling steps are used to optimize the point cloud with a parallel algorithm; and then the point cloud is reorganized according to the scene through the point cloud reorganization step. This algorithm process has significant advantages in the fields of robotics and surveying and mapping, which can greatly improve the quality of point clouds, enhance the actual application effect, and provide an efficient, general and easy-to-implement strategy for data processing in related industries, which will effectively promote the development of point cloud technology and the deepening of practice.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method and device for point cloud data processing and refined calculation, and a readable storage medium thereof. Background Art

[0002] In recent years, with the continuous advancement of sensor technology and terminal computing power, real-time positioning and mapping technology has achieved rapid development in the robotics and surveying industries. However, due to the limitations of sensor accuracy and current algorithm accuracy, the generated point clouds often have many problems, mainly reflected in low accuracy, insufficient refinement, and obvious noise. These problems greatly limit the effective use of point clouds in actual production and life.

[0003] In the field of robotics, insufficient point cloud accuracy may lead to deviations in robot navigation and positioning, making it impossible to accurately identify and avoid obstacles, affecting its operational safety and task execution efficiency. For example, in autonomous navigation in complex environments, inaccurate point cloud data may cause errors in the robot's perception of the surrounding environment, leading to incorrect decisions.

[0004] In the surveying and mapping industry, insufficient refinement of point clouds will affect the quality of surveying and mapping results and fail to meet high-precision surveying and mapping requirements, such as terrain surveying and building modeling. Obvious noise may interfere with the extraction of real terrain and object features, reducing the reliability of surveying and mapping data.

[0005] Therefore, it is particularly important to develop an efficient, universal and easy-to-implement point cloud data processing and refinement calculation. This invention proposes a point cloud data processing and refinement algorithm to solve these problems. Through innovative iterative segmentation and interpolation technology and other means, it is committed to improving the accuracy and refinement level of point clouds, reducing the impact of noise, and thus expanding the application potential of point clouds in various fields. Summary of the invention

[0006] The embodiments of the present invention provide a point cloud data processing and refined calculation method, device and readable storage medium thereof, which are aimed at solving the problems that the current technology is limited by sensor accuracy and algorithm accuracy, resulting in the generated point cloud having low accuracy, insufficient refinement, obvious noise points, etc., which limits its application in actual production and life.

[0007] The core technology of the present invention is mainly a kind of point cloud data processing and refined calculation, which covers the steps of noise removal, point cloud segmentation and recognition, linear fitting, segmentation merging and resampling, point cloud reorganization and other steps. It improves the accuracy and refinement level of point cloud through iterative segmentation interpolation technology and a variety of innovative algorithms, removes noise and reorganizes point cloud for easy use.

[0008] In a first aspect, the present invention provides a point cloud data processing and refined calculation method, the method comprising the following steps:

[0009] S00, identify and remove noise points in point cloud data;

[0010] S10, calculating a normal vector by local normal vector calculation, based on the calculated normal vector;

[0011] S20, removing noise points with consistency of local normal vectors, and eliminating points and their normal vectors with a value greater than a set threshold;

[0012] S30, based on the normal vector smoothness, the custom feature growth segmentation process traverses the point cloud, performs the custom feature growth according to the normal vector consistency and smoothness, and segments the point clouds in different directions;

[0013] S40, performing linear fitting on the segmented point cloud, and screening, segmenting and merging the fitting results, and then performing point cloud projection and resampling to obtain a refined point cloud;

[0014] S50, reorganizing the refined point cloud and the remaining point clouds by weight and label according to the usage scenario, and saving the final refined point cloud;

[0015] The remaining point cloud is the point set that does not participate in fitting or fails in fitting segmentation and merging.

[0016] Furthermore, in step S00, based on the expansion grid algorithm, the grid map is first read and expanded according to the expansion coefficient and expansion distance, and then the grid is traversed. If the grid occupancy value is greater than the threshold, it is identified as an obstacle, and the grid in the neighborhood is expanded and updated to provide a basic range for subsequent accurate judgment of noise points.

[0017] Furthermore, in step S00, a noise removal algorithm based on custom feature growth judgment is used to read point cloud data, and noise points are initially determined according to the grid values ​​corresponding to the point cloud. Custom feature growth is performed on the initially determined noise points, and the grid values ​​corresponding to the point cloud after growth are statistically counted, and the average grid occupancy value of the point cloud and the point cloud covariance are calculated. Based on this, it is determined whether the point cloud is a noise point and is removed to improve the accuracy of noise removal.

[0018] Furthermore, in step S10, a local normal vector is calculated, a kd-tree is constructed, a binary search is performed through the kd-tree, the point cloud data is traversed to find the nearest neighbor point, a straight line is fitted to the nearest neighbor point found, and its normal vector is calculated based on the fitted straight line.

[0019] Furthermore, the specific steps of step S20 are:

[0020] Traverse the point cloud data, search for the nearest neighbor, and calculate the consistency of the normal vector between the current point and the neighboring point;

[0021] If it is less than the threshold, it is retained;

[0022] If it is greater than the threshold, the point and its normal vector are removed.

[0023] Furthermore, in step S40, the linear fitting step is:

[0024] Fitting is performed based on the segmented single point cloud data, the direction vector of the fitted line is calculated, the point cloud is projected to the direction of the line, the starting point and end point of the line are calculated, and the mean Euclidean distance from the segmentation point to the line is calculated as the fitting error;

[0025] The steps of splitting and merging are:

[0026] The point cloud corresponding to the fitting line whose fitting error is greater than the threshold is segmented twice, and the point clouds with similar directions and starting or ending point distances less than the threshold are merged. The merged point cloud is re-fitted and segmented, and it is iterated until there is no new segmentation or merging or the number of iterations reaches the upper limit.

[0027] Furthermore, in step S40, the resampling step is:

[0028] The segmented point cloud is projected in the direction of the fitted line, and the final point cloud is retained according to the preset step size.

[0029] In a second aspect, the present invention provides a point cloud data processing and refinement computing device, comprising:

[0030] The noise removal module identifies and removes noise points in point cloud data based on the noise removal algorithm based on the expansion grid and custom feature growth discrimination; calculates the normal vector through local normal vector calculation, and removes points and their normal vectors that are greater than the set threshold through local normal vector consistency noise removal;

[0031] Point cloud segmentation module, based on the custom feature growth segmentation processing of normal vector smoothness, traverses the point cloud, performs custom feature growth according to the consistency and smoothness of the normal vector, and segments point clouds in different directions;

[0032] The linear fitting, segmentation, merging and resampling module performs linear fitting on the segmented point cloud, and screens, segmentates and merges the fitting results, and then performs point cloud projection and resampling to obtain a refined point cloud.

[0033] The point cloud reorganization module reorganizes the refined point cloud and the remaining point cloud by weight and label according to the usage scenario, and saves the final refined point cloud; the remaining point cloud is the point set that does not participate in fitting or fails to fit, split and merge.

[0034] In a third aspect, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the above-mentioned point cloud data processing and refined calculation method.

[0035] In a fourth aspect, the present invention provides a readable storage medium, in which a computer program is stored. The computer program includes a program code for controlling a process to execute the process, and the process includes the point cloud data processing and refined calculation method described above.

[0036] The main contributions and innovations of the present invention are as follows:

[0037] 1. Improve point cloud accuracy: The present invention adopts iterative segmentation and interpolation technology to adaptively process point clouds through multiple iterations. In the point cloud segmentation and recognition step, the local normal vector is accurately calculated, and the point cloud is accurately divided by removing the noise points of the local normal vector consistency and the custom feature growth segmentation processing based on the normal vector smoothness. The linear fitting, segmentation merging and resampling steps are further optimized to improve the overall accuracy, far exceeding the limitations of existing technologies, and providing a reliable data foundation for robot navigation, surveying and mapping modeling, etc. For example, the terrain details in surveying and mapping are restored more accurately, which helps to build high-precision maps and models.

[0038] 2. Effectively remove noise: The noise removal algorithm based on expanded grid and custom feature growth discrimination has unique advantages. By reading the grid map expansion process, combining point cloud traversal and feature growth discrimination, the noise characteristics are considered in multiple dimensions. Unlike existing technologies that only rely on a single threshold or simple filtering, this algorithm comprehensively analyzes the spatial distribution and feature association of the point cloud, accurately removes noise, avoids accidental deletion of valid points, ensures the purity of point cloud data, improves data quality and subsequent processing accuracy, and has strong stability and high reliability in point cloud processing in complex environments.

[0039] 3. Maintaining geometric information and ease of use: The point cloud reconstruction process is a key link in the present invention. The refined point cloud after sampling and the remaining point cloud are reorganized by weight and label according to the usage scenario, which not only retains the local geometric information of point clouds with different attributes, but also takes into account the differences in application requirements. This is different from the existing technology that ignores information integrity or ease of use. It makes point clouds more flexible and efficient in multiple scenarios such as target detection and object recognition, accelerates the data processing process, improves the applicability of the algorithm, and provides a universal and adaptable point cloud processing solution for various fields, promoting industry development and innovative application expansion.

[0040] The details of one or more embodiments of the invention are set forth in the following drawings and description so that other features, objects, and advantages of the invention are more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0042] Figure 1This is a flow chart of a point cloud data processing and refined calculation method according to an embodiment of the present invention;

[0043] Figure 2 4 is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with one or more embodiments of this specification. Instead, they are merely examples of devices and methods consistent with some aspects of one or more embodiments of this specification as detailed in the appended claims.

[0045] It should be noted that: in other embodiments, the steps of the corresponding method are not necessarily performed in the order shown and described in this specification. In some other embodiments, the steps included in the method may be more or less than those described in this specification. In addition, a single step described in this specification may be decomposed into multiple steps for description in other embodiments; and multiple steps described in this specification may be combined into a single step for description in other embodiments.

[0046] The current technology is limited by the accuracy of sensors and algorithms, resulting in the generated point cloud having problems such as low accuracy, insufficient refinement, and obvious noise, which restricts its application in actual production and life.

[0047] Based on this, the present invention solves the problems existing in the prior art based on iterative segmentation interpolation and the like.

[0048] Embodiment 1

[0049] The present invention aims to propose a point cloud data processing and refined calculation method, specifically, referring to Figure 1 , the method comprises the following steps:

[0050] S1. Obtain original point cloud data and raster map data through surveying and mapping software or mapping algorithms.

[0051] S2, noise removal step, reads point cloud data and raster map data, and uses a noise removal algorithm based on expansion grid and custom feature growth discrimination to identify and remove noise in the point cloud.

[0052] In this embodiment, the grid map reading and expansion includes reading a grid map, denoted as G, and expanding the grid map according to an expansion coefficient λ and an expansion distance d. The expansion method is:

[0053] Traverse the grid, if the grid occupancy value is greater than the threshold, it is identified as an obstacle, and the grid in the neighborhood is expanded and updated according to the expansion coefficient and expansion distance. The expansion distance and expansion coefficient are determined by the robot's circumscribed circle radius and the grid attenuation rate when the grid map is created. The larger the expansion distance, the lower the attenuation rate, and the greater the noise removal.

[0054] The expansion distance and expansion coefficient are determined by the radius of the robot's circumscribed circle and the grid attenuation rate when the grid map is created. The larger the expansion distance, the lower the attenuation rate, and the greater the noise removal. The expansion method is to traverse the grid. If the grid occupancy value is greater than the threshold, it is identified as an obstacle. The grid in the neighborhood is expanded and updated according to the expansion coefficient and expansion distance. The update formula is as follows:

[0055]

[0056] in, is the domain grid occupancy value, is the current grid occupancy value, It is the Euclidean distance from the center of the domain grid to the current grid.

[0057] Among them, the noise removal algorithm based on expansion grid and custom feature growth discrimination is:

[0058] According to the corresponding grid value of the point cloud, it is initially determined whether it is a noise point. If it is, custom feature growth is performed to obtain the point cloud after growth. The corresponding grid value of the point cloud after growth is counted, the average grid occupancy value of the point is calculated, and the point cloud covariance is counted. According to the average grid occupancy value and point cloud covariance, it is determined whether the point cloud is a noise point. If it is determined to be a noise point, it is removed. The specific steps are as follows:

[0059] The original point cloud is called P, which consists of n points. is 0( Represents the three-dimensional coordinates in the map coordinate system, taking the plane point cloud as an example, so z i =0):

[0060]

[0061] Traverse the point cloud P, if Corresponding grid value , use this point as the initial point for custom feature growth; g t It is a preset threshold value, which can be set according to actual needs.

[0062] make , the nearest neighbor point is searched as ,if ,and , Euclidean distance is less than the growth distance threshold , update the search point to ,Will Place in growing container ; Repeat the above steps until there is no point that meets the conditions, and the custom feature growth ends, and we get:

[0063]

[0064] Calculate the mean, covariance, and average occupancy value of the growth point cloud. The calculation formula is as follows:

[0065]

[0066]

[0067]

[0068] Where m represents the growth container The number of points in ; Growing container The covariance of the points in ; Represents a growth container The average occupancy value of the points in .

[0069] right Perform SVD (Singular Value Decomposition, an important matrix decomposition method) decomposition and calculate the eigenvalue ratio ,and ,but For noise, delete the noise in P. Delete noise Red is the recognized noise and will be deleted.

[0070] in and express The maximum and minimum eigenvalues ​​obtained by SVD decomposition, is the ratio of the maximum to the minimum eigenvalue, is the eigenvalue ratio threshold, is the average grid occupancy threshold.

[0071] In the point cloud data processing of the present invention, SVD decomposition is used in multiple key steps. For example, in the point cloud traversal of noise removal and the noise removal based on custom feature growth, the growing point cloud is decomposed by SVD, and the eigenvalue ratio is calculated to determine whether the point cloud is a noise point. In the local normal vector calculation, the neighboring point set is decomposed by SVD to obtain the normal vector of the target point. Through SVD decomposition, key feature information can be extracted from complex point cloud data relationships, such as using the size of singular values ​​to reflect the degree of local change of the point cloud, and determining the direction of geometric features according to the direction of the singular vector, thereby achieving accurate noise point identification, reliable normal vector calculation and efficient point cloud segmentation, improving the accuracy and quality of point cloud processing, and laying the foundation for subsequent refined processing.

[0072] S3, point cloud segmentation and recognition step, accurately segment the point cloud through local normal vector calculation, local normal vector consistency noise removal, and custom feature growth segmentation processing based on normal vector smoothness.

[0073] In this embodiment, the point cloud segmentation and recognition steps include local normal vector calculation, local normal vector consistency noise removal, and custom feature growth segmentation processing based on normal vector smoothness to accurately segment the point cloud.

[0074] The local normal vector calculation further includes constructing a kd-tree (short for k-dimensional tree, which is a tree data structure that stores instance points in k-dimensional space for fast retrieval. It is a special binary space partition tree, where each node represents a spatial range), traversing the point cloud to find the nearest neighbor, fitting the nearest point straight line, and calculating the normal vector. The specific steps are as follows:

[0075] remember For the current point, find the nearest neighbor point in the kd-tree and record it as ,

[0076]

[0077] Calculate the mean and covariance of the growing point cloud. The calculation formula is as follows:

[0078]

[0079]

[0080] right Perform SVD decomposition, and the characteristic vector corresponding to the minimum eigenvalue is The normal vector .

[0081] Among them, local normal vector consistency noise removal is achieved by traversing the point cloud, searching for the nearest neighbor, calculating the consistency of the normal vectors of the current point and the neighboring points. If the normal vector is less than the threshold, it is retained, and if the normal vector is greater than the threshold, the point and its normal vector are removed. The normal vector consistency calculation formula is as follows:

[0082]

[0083] Among them, the custom feature growth segmentation process based on normal vector smoothness is:

[0084] Traverse the point cloud, perform custom feature growth based on the consistency and smoothness of the normal vector, and segment point clouds in different directions. The custom feature growth process is the same as in step a. The normal vector smoothness calculation method is as follows:

[0085]

[0086] Right now Normal vector smoothing consistency judgment threshold. Normal vectors with an angle greater than this threshold will be removed. is the cosine of the angle threshold.

[0087] The set of points after growth segmentation that meets the conditions is denoted as S:

[0088]

[0089] In the segmentation instance, different colors can be used to represent different cluster segmentation point clouds.

[0090] S4, linear fitting, segmentation, merging and resampling steps, through a parallelized linear fitting algorithm, efficiently fit the point cloud data, and screen, segment and merge the fitting results, and finally perform point cloud projection and resampling to obtain a refined point cloud.

[0091] In this embodiment, linear fitting is performed based on the segmented single point cloud data, the direction vector of the fitted line is calculated, the point cloud is projected to the direction of the line, and the starting point of the line is calculated. ,end , calculate the mean of the Euclidean distance from the segmentation point to the straight line as the fitting error e. The specific algorithm steps are as follows:

[0092] For each point set Perform straight line fitting and get the preliminary fitting results ,in , , d, and e represent the starting point, end point, direction vector, and fitting error of the fitting line, respectively.

[0093]

[0094] Fit all point cloud sets to get the corresponding line set L:

[0095]

[0096] The segmentation and merging further includes the large error (greater than the threshold e t ) is used to perform secondary segmentation on the point cloud corresponding to the fitting line, and the segmentation result is fitted again (the fitting operation is consistent with the previous fitting step). The point clouds with similar directions and starting point or end point distance less than the threshold are merged, and the segmented and merged point cloud is refitted. It is iterated until no segmentation or merging is required or the specified number of times is reached to exit, and the final point set and its corresponding line set S', L' are obtained.

[0097] Resampling includes projecting the segmented point cloud in the direction of the fitted line and retaining the final point cloud at a certain step size. Project the points in the point set L' onto the corresponding line and retain them according to the resolution to obtain the resampled points L''. In the final segmentation point set, green can be represented as resampled points, and other color points are represented as retained points.

[0098] S5, point cloud reorganization step, reorganizes the sampled refined point cloud and the remaining point cloud by weight and label according to the usage scenario, and saves the final refined point cloud.

[0099] In this embodiment, the point cloud reorganization step includes point cloud attribute reorganization, classification by resampled point L'' and other points, and different weights are assigned according to different categories. Other points are point sets that do not participate in straight line fitting or fail to segment and merge straight line fitting, and the final point set is saved.

[0100] Embodiment 2

[0101] Based on the same concept, the present invention also proposes a point cloud data processing and refined calculation device, comprising:

[0102] The noise removal module identifies and removes noise points in point cloud data based on the noise removal algorithm based on the expansion grid and custom feature growth discrimination; calculates the normal vector through local normal vector calculation, and removes points and their normal vectors that are greater than the set threshold through local normal vector consistency noise removal;

[0103] Point cloud segmentation module, based on the custom feature growth segmentation processing of normal vector smoothness, traverses the point cloud, performs custom feature growth according to the consistency and smoothness of the normal vector, and segments point clouds in different directions;

[0104] The linear fitting, segmentation, merging and resampling module performs linear fitting on the segmented point cloud, and screens, segmentates and merges the fitting results, and then performs point cloud projection and resampling to obtain a refined point cloud.

[0105] The point cloud reorganization module reorganizes the refined point cloud and the remaining point cloud by weight and label according to the usage scenario, and saves the final refined point cloud; the remaining point cloud is the point set that does not participate in fitting or fails to fit, split and merge.

[0106] Embodiment 3

[0107] This embodiment also provides an electronic device, referring to Figure 2 , comprises a memory 404 and a processor 402, wherein the memory 404 stores a computer program, and the processor 402 is configured to run the computer program to execute the steps in any of the above method embodiments.

[0108] Specifically, the processor 402 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiment of the present invention.

[0109] Among them, the memory 404 may include a large capacity memory 404 for data or instructions. For example, but not limitation, the memory 404 may include a hard disk drive (HDD), a floppy disk drive, a solid state drive (SSD), a flash memory, an optical disk, a magneto-optical disk, a tape, or a universal serial bus (USB) drive, or a combination of two or more of these. In appropriate cases, the memory 404 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 404 may be inside or outside the data processing device. In a specific embodiment, the memory 404 is a non-volatile memory. In a specific embodiment, the memory 404 includes a read-only memory (ROM) and a random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (Programmable Read-Only Memory, PROM for short), an erasable PROM (Erasable Programmable Read-Only Memory, EPROM for short), an electrically erasable PROM (Electrically Erasable Programmable Read-Only Memory, EEPROM for short), an electrically alterable ROM (Electrically Alterable Read-Only Memory, EAROM for short) or a flash memory (FLASH) or a combination of two or more of these. In appropriate circumstances, the RAM may be a static random access memory (SRAM) or a dynamic random access memory (DRAM), wherein the DRAM may be a fast page mode dynamic random access memory 404 (FPMDRAM), an extended data output dynamic random access memory (EDODRAM), a synchronous dynamic random access memory (SDRAM), etc.

[0110] The memory 404 may be used to store or cache various data files required for processing and / or communication, as well as possible computer program instructions executed by the processor 402 .

[0111] The processor 402 reads and executes computer program instructions stored in the memory 404 to implement any one of the point cloud data processing and refined calculation methods in the above embodiments.

[0112] Optionally, the electronic device may further include a transmission device 406 and an input / output device 408 , wherein the transmission device 406 is connected to the processor 402 , and the input / output device 408 is connected to the processor 402 .

[0113] The transmission device 406 can be used to receive or send data via a network. Specific examples of the above-mentioned network may include a wired or wireless network provided by a communication provider of the electronic device. In one example, the transmission device includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 406 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0114] Input / output devices 408 are used to input or output information.

[0115] Embodiment 4

[0116] This embodiment also provides a readable storage medium, in which a computer program is stored. The computer program includes a program code for controlling a process to execute a process. The process includes the point cloud data processing and refined calculation method according to the first embodiment.

[0117] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0118] In general, various embodiments may be implemented in hardware or dedicated circuits, software, logic, or any combination thereof. Some aspects of the invention may be implemented in hardware, while other aspects may be implemented in firmware or software executed by a controller, microprocessor, or other computing device, but the invention is not limited thereto. Although various aspects of the invention may be shown and described as block diagrams, flow charts, or using some other graphical representation, it should be understood that, as non-limiting examples, the boxes, devices, systems, techniques, or methods described herein may be implemented in hardware, software, firmware, dedicated circuits or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0119] Embodiments of the present invention may be implemented by computer software that is executable by a data processor of a mobile device, such as in a processor entity, or by hardware, or by a combination of software and hardware. Computer software or programs (also referred to as program products) including software routines, applets and / or macros may be stored in any device-readable data storage medium, and they include program instructions for performing specific tasks. A computer program product may include one or more computer executable components configured to perform an embodiment when the program is run. One or more computer executable components may be at least one software code or a portion thereof. In addition, at this point, it should be noted that, for example, Figure 1 Any block of the logic flow in the program may represent program steps, or interconnected logic circuits, blocks and functions, or a combination of program steps and logic circuits, blocks and functions. The software may be stored on physical media such as memory chips or storage blocks implemented within a processor, magnetic media such as hard disks or floppy disks, and optical media such as, for example, DVDs and their data variants, CDs, etc. Physical media are non-transitory media.

[0120] Those skilled in the art should understand that the technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0121] The above embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present invention. It should be pointed out that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the attached claims.

Claims

1. A point cloud data processing and refined calculation method, characterized in that: The following steps are involved: S00, identify and remove noise points in point cloud data; S10, calculating a normal vector by local normal vector calculation, based on the calculated normal vector; S20, removing noise points with consistency of local normal vectors, and eliminating points and their normal vectors with a value greater than a set threshold; S30, based on the normal vector smoothness, the custom feature growth segmentation process traverses the point cloud, performs the custom feature growth according to the normal vector consistency and smoothness, and segments the point clouds in different directions; S40, performing linear fitting on the segmented point cloud, and screening, segmenting and merging the fitting results, and then performing point cloud projection and resampling to obtain a refined point cloud; The steps of the linear fitting are: Fitting is performed based on the segmented single point cloud data, the direction vector of the fitted line is calculated, the point cloud is projected to the direction of the line, the starting point and end point of the line are calculated, and the mean Euclidean distance from the segmentation point to the line is calculated as the fitting error; The steps of segmentation and merging are: Perform secondary segmentation on the point cloud corresponding to the fitting line whose fitting error is greater than the first threshold, merge the point clouds whose directions are close and whose distance from the starting point or the end point of the fitting line is less than the second threshold, re-fit the segmented and merged point cloud, and iterate until there is no new segmentation or merging or the number of iterations reaches the upper limit; The resampling steps are: Project the segmented point cloud in the direction of the fitted line and retain the final point cloud according to the preset step size; S50, reorganizing the refined point cloud and the remaining point clouds by weight and label according to the usage scenario, and saving the final refined point cloud; The remaining point cloud is the point set that does not participate in fitting or fails in fitting segmentation and merging.

2. A point cloud data processing and refined calculation method as claimed in claim 1, characterized in that: In step S00, based on the expansion grid algorithm, the grid map is first read and expanded according to the expansion coefficient and expansion distance, and then the grid is traversed. If the grid occupancy value is greater than the threshold, it is identified as an obstacle, and the grid in the neighborhood is expanded and updated to provide a basic range for subsequent accurate judgment of noise points.

3. A point cloud data processing and refined calculation method as claimed in claim 2, characterized in that: In step S00, a noise removal algorithm based on custom feature growth judgment is used to read point cloud data, preliminarily determine noise points according to the grid values ​​corresponding to the point cloud, perform custom feature growth on the preliminarily determined noise points, count the grid values ​​corresponding to the point cloud after growth, calculate the average grid occupancy value of the point cloud and the point cloud covariance, and determine whether the point cloud is a noise point based on this and remove it to improve the accuracy of noise removal.

4. A point cloud data processing and refined calculation method as claimed in claim 1, characterized in that: In step S10, the local normal vector is calculated, a kd-tree is constructed, a binary search is performed through the kd-tree, the point cloud data is traversed to find the nearest neighbor point, a straight line is fitted to the nearest neighbor point found, and its normal vector is calculated based on the fitted straight line.

5. The point cloud data processing and refined calculation method according to claim 1, characterized in that: The specific steps of step S20 are: Traverse the point cloud data, search for the nearest neighbor, and calculate the consistency of the normal vector between the current point and the neighboring point; If it is less than the threshold, it is retained; If it is greater than the threshold, the point and its normal vector are removed.

6. A point cloud data processing and refined calculation device, characterized in that: include: Noise removal module, based on the noise removal algorithm of expansion grid and custom feature growth discrimination, identifies and removes noise in point cloud data; Calculate the normal vector through local normal vector calculation, based on the calculated normal vector; remove the noise point through local normal vector consistency, and remove the points and their normal vectors that are greater than the set threshold; Point cloud segmentation module, based on the custom feature growth segmentation processing of normal vector smoothness, traverses the point cloud, performs custom feature growth according to the consistency and smoothness of the normal vector, and segments point clouds in different directions; The linear fitting, segmentation, merging and resampling module performs linear fitting on the segmented point cloud, and screens, segmentates and merges the fitting results, and then performs point cloud projection and resampling to obtain a refined point cloud. Wherein, the steps of linear fitting are: Fitting is performed based on the segmented single point cloud data, the direction vector of the fitted line is calculated, the point cloud is projected to the direction of the line, the starting point and end point of the line are calculated, and the mean Euclidean distance from the segmentation point to the line is calculated as the fitting error; The steps of segmentation and merging are: Perform secondary segmentation on the point cloud corresponding to the fitting line whose fitting error is greater than the first threshold, merge the point clouds whose directions are close and whose distance from the starting point or the end point of the fitting line is less than the second threshold, re-fit the segmented and merged point cloud, and iterate until there is no new segmentation or merging or the number of iterations reaches the upper limit; The resampling steps are: Project the segmented point cloud in the direction of the fitted line and retain the final point cloud according to the preset step size; The point cloud reorganization module reorganizes the refined point cloud and the remaining point cloud by weight and label according to the usage scenario, and saves the final refined point cloud; the remaining point cloud is the point set that does not participate in fitting or fails to fit, split and merge.

7. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the point cloud data processing and refined calculation method according to any one of claims 1 to 5.

8. A readable storage medium, characterized in that: The readable storage medium stores a computer program, which includes a program code for controlling a process to execute a process, and the process includes the point cloud data processing and refined calculation method according to any one of claims 1 to 5.

Images