Coal mine three-dimensional point cloud registration and fixed facility removal method in warehouse environment

By using high-precision lidar and regional covariance characteristics RANSAC registration algorithm, ICP algorithm and plane segmentation algorithm in warehouse environment, the problems of three-dimensional point cloud registration and fixed facilities removal of coal mines are solved, and efficient and accurate points are achieved Cloud data processing.

CN118982567APending Publication Date: 2024-11-19SHAANXI LOGISTICS GRP IND RES INST CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411073228.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

In the prior art, it is difficult to perform coarse and fine registration of coal mine three-dimensional point clouds based on fixed camera poses in warehouse environments, especially when using ICP algorithms, there are problems of insufficient computational complexity and accuracy.

Method used

Multiple high-precision lidar equipment are used to scan coal mine areas, and the covariance characteristics of RANSAC registration algorithm and ICP algorithm are used to register coarseness, and the point cloud data is divided by combining the plane segmentation algorithm. The data points are judged by normal vector projection. Attributes, downsampling processing to reduce the amount of calculation.

Benefits of technology

It realizes more efficient point cloud data registration and segmentation, reduces computing complexity, improves the accuracy and speed of point cloud data processing, and is suitable for real-time registration tasks in complex scenarios.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118982567A_ABST
    Figure CN118982567A_ABST
Patent Text Reader

Abstract

The invention discloses a coal mine three-dimensional point cloud registration and fixed facility removal method in a warehouse environment, relates to the technical field of three-dimensional point cloud registration and point cloud segmentation, and solves the problem that in the prior art, rough registration based on the pose of a fixed camera and fine registration based on ICP cannot be achieved. The method comprises the steps that a plurality of high-precision laser radar devices are adopted to carry out scanning and preprocessing on a coal mine area, then coordinate system conversion is carried out, fusion is carried out, and coarse alignment discrete point cloud data are obtained; fine registration is carried out by using the optimal rotation matrix to obtain fine registration discrete point cloud data; segmenting the fine registration discrete point cloud data, calculating a normal vector of each data point in the non-planar point cloud data in a preset range, obtaining a projection value, determining an attribute of each data point, and obtaining a clustering result; according to the method, the advantages of combining local and global information, reducing the calculation complexity and the like are realized, and the segmentation task of the point cloud data can be better processed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of three-dimensional point cloud registration and point cloud segmentation, and in particular to a method for three-dimensional point cloud registration and fixed facility removal in a warehouse environment. Background Art

[0002] The main idea of ​​the 3D point cloud registration process is to convert two or more pieces of point cloud data into the same coordinate system through a registration algorithm to obtain a complete 3D model of the object. At present, point cloud registration methods can be divided into three categories: registration methods based on geometric features, registration methods based on iterations, and registration methods based on surfaces. 3D point cloud registration is a key research issue in computer vision and currently has important application value in technical fields such as pose estimation, reverse engineering, SLAM, and image processing.

[0003] Point cloud building segmentation technology refers to the technology of segmenting and identifying buildings and other scenes in 3D point cloud data obtained by laser scanning and other technologies. This technology has important applications in urban planning, map making, virtual reality and augmented reality.

[0004] How to perform coarse registration based on fixed camera pose and fine registration based on ICP is an urgent problem to be solved. Summary of the invention

[0005] The present invention provides a method for coal mine three-dimensional point cloud registration and fixed facility removal in a warehouse environment, which solves the problem that the prior art cannot perform coarse registration based on fixed camera posture and fine registration based on ICP, and achieves the advantages of combining local and global information and reducing computational complexity. It can better handle the segmentation task of point cloud data and show superior performance and effects in many application scenarios.

[0006] In a first aspect, the present invention provides a method for coal mine three-dimensional point cloud registration and fixed facility removal in a warehouse environment, the method comprising:

[0007] Use multiple high-precision LiDAR devices to scan the coal mine area and obtain multiple discrete point cloud data sets;

[0008] Preprocessing the multiple discrete point cloud data sets to obtain multiple sampled discrete point cloud data sets, and determining a first sampled discrete point cloud data set and a second sampled discrete point cloud data set corresponding to the laser radar device;

[0009] According to the RANSAC registration algorithm and the ICP algorithm of the regional covariance feature, the first sampled discrete point cloud data set and the second sampled discrete point cloud data set are calculated to obtain a first translation matrix and a first rotation matrix, and a rough registration is performed according to the first translation matrix and the first rotation matrix to obtain rough aligned discrete point cloud data, wherein the rough aligned discrete point cloud data includes: a first point set and a second point set;

[0010] Using the ICP algorithm to calculate the coarsely aligned discrete point cloud data to obtain an optimal translation matrix and an optimal rotation matrix, and using the optimal rotation matrix to perform fine alignment to obtain finely aligned discrete point cloud data;

[0011] Segmenting the finely registered discrete point cloud data using a plane segmentation algorithm to obtain plane point cloud data and non-plane point cloud data;

[0012] Calculate the normal vector of each data point in the non-planar point cloud data within a preset range, project the normal vector onto the Z axis in the three-dimensional coordinate axis to obtain a projection value, and determine the attributes of the data point based on the projection value; wherein the attributes include: plane points and elevation points.

[0013] In combination with the first aspect, in a possible implementation manner, preprocessing the plurality of discrete point cloud data to obtain a first sampled discrete point cloud data set includes:

[0014] Filtering the plurality of discrete point cloud data according to a first threshold value to obtain a filtered discrete point cloud data set;

[0015] Downsampling is performed on the screened discrete point cloud data to obtain a first sampled discrete point cloud data set.

[0016] In combination with the first aspect, in a possible implementation manner, the RANSAC registration algorithm and the ICP algorithm based on the regional covariance feature calculate the first sampled discrete point cloud data set and the second sampled discrete point cloud data set to obtain a first translation matrix and a first rotation matrix, including:

[0017] Respectively calculating the covariance matrix corresponding to each discrete point cloud data in the first sampled discrete point cloud data set and the second sampled discrete point cloud data set;

[0018] Calculate the distance between the covariance matrix corresponding to each discrete point cloud data in the first sampled discrete point cloud data set and the covariance matrix corresponding to each discrete point cloud data in the second sampled discrete point cloud data set according to the distance calculation formula; obtain the distance set corresponding to each discrete point cloud data in the first sampled discrete point cloud data set;

[0019] Selecting discrete point cloud data in the second sampled discrete point cloud data set and discrete point cloud data in the first sampled discrete point cloud data set corresponding to the minimum distance in the distance set to form a matching point pair;

[0020] A plurality of matching point pairs are determined, and a first translation matrix and a first rotation matrix are obtained by calculating the plurality of matching point pairs using an ICP algorithm.

[0021] In combination with the first aspect, in a possible implementation manner, the covariance matrix is ​​expressed as:

[0022]

[0023] in, is any point P in the first sampled discrete point cloud data set and the second sampled discrete point cloud data set i The three-dimensional centroid of the K nearest neighbor set, λ j represents the jth eigenvalue corresponding to the covariance matrix; It represents the jth eigenvector corresponding to the covariance matrix; k represents the number of adjacent points.

[0024] In combination with the first aspect, in a possible implementation manner, the distance calculation formula is expressed as:

[0025]

[0026] Where C1 represents a covariance matrix for one data point in the first sampled discrete point cloud data set; C2 represents a covariance matrix for one data point in the second sampled discrete point cloud data set; λ i represents the generalized eigenvalue; m represents the number of eigenvalues.

[0027] In combination with the first aspect, in a possible implementation manner, the using of the ICP algorithm to calculate the coarsely aligned discrete point cloud data to obtain an optimal translation matrix and an optimal rotation matrix includes:

[0028] Determine a first optimal transformation formula, perform a first-order derivative on the first optimal transformation formula, and set the derivative to 0 to obtain an optimal translation matrix;

[0029] Without considering the translation error, a second optimal transformation formula is determined, and SVD decomposition calculation is performed on the second optimal transformation formula to obtain an optimal rotation matrix.

[0030] In combination with the first aspect, in a possible implementation manner, the first optimal transformation formula is expressed as:

[0031]

[0032] Among them, P tiIndicates that the distance between the data point in the first point set and the data point in the second point set is less than the preset threshold; p si' represents the coordinates of each point in the second point set after the initial change is applied; R represents the rotation matrix; t represents the translation matrix; n represents the number of data points whose distance between the data points in the first point set and the data points in the second point set is less than a preset threshold.

[0033] In combination with the first aspect, in a possible implementation manner, the second optimal transformation formula is expressed as:

[0034]

[0035] Among them, P ti The coordinates of the point in the first point set that is closest to the second point set; P si represents the data points whose distance between the data points in the first point set and the data points in the second point set is less than the preset threshold; R represents the rotation matrix; n represents the number of data points whose distance between the data points in the first point set and the data points in the second point set is less than the preset threshold.

[0036] In a second aspect, the present invention provides a coal mine three-dimensional point cloud registration and fixed facility removal device in a warehouse environment, the device comprising:

[0037] The data acquisition module is used to use multiple high-precision laser radar devices to scan the coal mine area and obtain multiple discrete point cloud data sets;

[0038] A sampling module, used to preprocess the multiple discrete point cloud data sets to obtain multiple sampled discrete point cloud data sets, and determine a first sampled discrete point cloud data set and a second sampled discrete point cloud data set corresponding to the laser radar device;

[0039] A coarse fusion module is used to calculate the first sampled discrete point cloud data set and the second sampled discrete point cloud data set according to the RANSAC registration algorithm of regional covariance features to obtain a first translation matrix and a first rotation matrix, and perform coarse registration according to the first translation matrix and the first rotation matrix to obtain coarsely aligned discrete point cloud data, wherein the coarsely aligned discrete point cloud data includes: a first point set and a second point set;

[0040] A fine fusion module is used to calculate the coarsely aligned discrete point cloud data using an ICP algorithm to obtain an optimal translation matrix and an optimal rotation matrix, and to perform fine registration using the optimal rotation matrix to obtain finely registered discrete point cloud data;

[0041] A data segmentation module, used to segment the finely registered discrete point cloud data using a plane segmentation algorithm to obtain plane point cloud data and non-plane point cloud data;

[0042] A data partitioning module is used to calculate the normal vector of each data point in the non-planar point cloud data within a preset range, and project the normal vector onto the Z axis in the three-dimensional coordinate axis to obtain a projection value, and determine the attributes of the data point based on the projection value; wherein the attributes include: plane points and facade points.

[0043] In a third aspect, the present invention provides a computer-readable storage medium having executable instructions, and when a computer executes the executable instructions, a method for three-dimensional point cloud registration and fixed facility removal of a coal mine in a warehouse environment can be implemented.

[0044] One or more technical solutions provided in the present invention have at least the following technical effects or advantages:

[0045] The present invention can obtain a large amount of discrete point cloud data containing coal mine information by utilizing high-precision laser radar scanning, providing a sufficient data basis for subsequent analysis; downsampling can effectively reduce the amount of calculation, making subsequent processing faster; since the on-site radar equipment is fixed, the point cloud can be roughly aligned according to the radar's position information, and the data can be preliminarily aligned at a rough level to reduce the overall error between the data and provide a starting point for subsequent more precise alignment; and the ICP method is used for alignment. The algorithm can achieve high-precision point cloud alignment under appropriate circumstances, accurately align two sets of point cloud data and find the best match between them; it shows a certain robustness when dealing with a certain degree of noise, occlusion or outliers, so that it can still produce good alignment results when facing complex point cloud data in the real world, and can usually converge quickly to the optimal solution, which is particularly suitable for scenarios that require real-time or fast alignment. The plane segmentation algorithm is widely applicable and fast; the building segmentation is based on the direction of the normal vector. This method is easy to implement and fast. Supervoxel segmentation is then used to cluster different coal mines. This method combines local and global information, and reduces computational complexity. It can better handle the segmentation task of point cloud data and demonstrate superior performance and effects in many application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A flowchart of the steps of a method for coal mine three-dimensional point cloud registration and fixed facility removal in a warehouse environment provided by an embodiment of the present invention;

[0047] Figure 2 A comparison diagram of point cloud registration before and after the laser radar provided in an embodiment of the present invention acquires point cloud data;

[0048] Figure 3 The rendering of the original marked building point cloud provided by the embodiment of the present invention;

[0049] Figure 4A visualization effect diagram after using supervoxel segmentation provided by an embodiment of the present invention;

[0050] Figure 5 This is a diagram showing the effect of the coal mine segmentation in the mining area of ​​the present invention provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] The present invention provides a method for coal mine three-dimensional point cloud registration and fixed facility removal in a warehouse environment. Figure 1 As shown, the method includes the following steps S101 to S106.

[0053] S101 uses multiple high-precision LiDAR devices to scan the coal mine area and obtain multiple discrete point cloud data sets.

[0054] For example, different high-precision laser radar devices are set at different angles to scan the coal mine area, generating a large amount of multiple discrete point cloud data including coal mine information.

[0055] S102, preprocessing multiple discrete point cloud data sets to obtain multiple sampled discrete point cloud data sets, and determining a first sampled discrete point cloud data set and a second sampled discrete point cloud data set corresponding to the laser radar device. Specifically, in step S102, preprocessing multiple discrete point cloud data sets to obtain a first sampled discrete point cloud data set includes the following steps.

[0056] (1) Filtering multiple discrete point cloud data according to a first threshold value to obtain a filtered discrete point cloud data set.

[0057] (2) Downsampling the filtered discrete point cloud data to obtain a first sampled discrete point cloud data set.

[0058] Exemplarily, the first threshold is set according to the computing power. The greater the computing power, the greater the first threshold, and the smaller the computing power, the smaller the first threshold.

[0059] S103, according to the RANSAC registration algorithm and ICP algorithm of regional covariance features, calculate the first sampled discrete point cloud data set and the second sampled discrete point cloud data set to obtain a first translation matrix and a first rotation matrix, and perform coarse registration according to the first translation matrix and the first rotation matrix to obtain coarsely aligned discrete point cloud data, wherein the coarsely aligned discrete point cloud data includes: a first point set and a second point set.

[0060] Specifically, in step S103, the first sampled discrete point cloud data set and the second sampled discrete point cloud data set are calculated according to the RANSAC registration algorithm and the ICP algorithm of the regional covariance feature to obtain a first translation matrix and a first rotation matrix, including steps S1031 to S1034.

[0061] S1031, respectively calculating the covariance matrix corresponding to each discrete point cloud data in the first sampled discrete point cloud data set and the second sampled discrete point cloud data set;

[0062] S1032, calculating the distance between the covariance matrix corresponding to each discrete point cloud data in the first sampled discrete point cloud data set and the covariance matrix corresponding to each discrete point cloud data in the second sampled discrete point cloud data set according to the distance calculation formula; obtaining a distance set corresponding to each discrete point cloud data in the first sampled discrete point cloud data set;

[0063] S1033, selecting discrete point cloud data in the second sampled discrete point cloud data set and discrete point cloud data in the first sampled discrete point cloud data set corresponding to the minimum distance in the distance set to form a matching point pair;

[0064] S1034, determining a plurality of matching point pairs, and calculating the plurality of matching point pairs using an ICP algorithm to obtain a first translation matrix and a first rotation matrix.

[0065] Specifically, the covariance matrix is ​​expressed as:

[0066]

[0067] in, is any point P in the first sampled discrete point cloud dataset and the second sampled discrete point cloud dataset i The three-dimensional centroid of the K nearest neighbor set, λ j represents the jth eigenvalue corresponding to the covariance matrix; It represents the jth eigenvector corresponding to the covariance matrix; k represents the number of adjacent points.

[0068] The distance calculation formula is expressed as:

[0069]

[0070] Where C1 represents a covariance matrix for one data point in the first sampled discrete point cloud data set; C2 represents a covariance matrix for one data point in the second sampled discrete point cloud data set; λ i represents the generalized eigenvalue; m represents the number of eigenvalues.

[0071] where {λ(C1,C2)} i=1...n is the generalized eigenvalue of matrices C1 and C2, which can be calculated as follows:

[0072] λ i C1x i -C2x i =0,i=1…d

[0073] d is the dimension of the covariance matrix, usually 3, x i ≠0 is a generalized eigenvector, so we need to calculate λ1(C1-C2),…,λ n The value of (C1-C2) requires a solution for λ such that det(λC1-C2)=0.

[0074] Simplifying the formula, we get:

[0075]

[0076] Since the positive definite real symmetric matrix The eigenvalues ​​of are positive real numbers, so for λ1(C1-C2),…,λ n The calculation of (C1-C2) is converted to Solving for the eigenvalues ​​of .

[0077] Assume that the positive definite matrix C1 -1 The square root of is B, then there exists an orthogonal matrix U, and U is processed as follows.

[0078]

[0079] Assumptions Then C1 -1 =UΛU'=UQ 2 U'

[0080] B 2 =C1 -1 , so we get B = UQU', solve the eigenvalues ​​of the matrix BC2B, and then calculate the distance.

[0081] Exemplarily, given an adjacent first sampled discrete point cloud data set S and a second sampled discrete point cloud data set M, the algorithm steps of the RANSAC registration algorithm based on regional covariance features are detailed as follows:

[0082] (1) Calculate the regional covariance characteristics of each point of the first sampled discrete point cloud dataset S and the second sampled discrete point cloud dataset M respectively.

[0083] (2) calculating the covariance matrix of each data point in the first sampled discrete point cloud data set S and the second sampled discrete point cloud data set M;

[0084] (3) respectively calculating the distance between the covariance matrix of each point in the first sampled discrete point cloud data set S and each point in the second sampled discrete point cloud data set M, selecting the smallest distance, and determining the corresponding point of each data point in the first sampled discrete point cloud data set S in the second sampled discrete point cloud data set M;

[0085] (4) randomly selecting multiple pairs of matching points and calculating the first translation matrix and the first rotation matrix;

[0086] (5) Using the first translation matrix and the first rotation matrix, the first sampled discrete point cloud data set S is processed to obtain coarsely aligned discrete point cloud data; wherein the coarsely aligned discrete point cloud data includes the first point set and the second point set.

[0087] S104, using the ICP algorithm to calculate the coarsely aligned discrete point cloud data to obtain an optimal translation matrix and an optimal rotation matrix, and using the optimal rotation matrix to perform fine alignment to obtain finely aligned discrete point cloud data.

[0088] Specifically, in step S104, the coarsely aligned discrete point cloud data is calculated using the ICP algorithm to obtain an optimal translation matrix and an optimal rotation matrix, including S1041 to S1042.

[0089] S1041, determining a first optimal transformation formula, taking a first-order derivative of the first optimal transformation formula, and setting the derivative to 0, to obtain an optimal translation matrix;

[0090] S1042, without considering the translation error, determine the second optimal transformation formula, perform SVD decomposition calculation on the second optimal transformation formula, and obtain the optimal rotation matrix.

[0091] like Figure 2 As shown in Figure 2, it is a comparison diagram before and after point cloud registration after the laser radar acquires point cloud data.

[0092] Specifically, the first optimal transformation formula is expressed as:

[0093]

[0094] Among them, P ti Indicates that the distance between the data point in the first point set and the data point in the second point set is less than the preset threshold; p si'represents the coordinates of each point in the second point set after the initial change is applied; R represents the rotation matrix; t represents the translation matrix; n represents the number of data points whose distance between the data points in the first point set and the data points in the second point set is less than a preset threshold.

[0095] Specifically, the formula for calculating the optimal translation matrix is ​​expressed as:

[0096]

[0097] Let the derivative be 0 and we get:

[0098]

[0099] Among them, P ti The coordinates of the point in the first point set that is closest to the second point set, P si represents a point in the origin cloud; represents the average coordinates of the target point; R represents the rotation matrix; Represents the average coordinate value of the source point cloud; n represents the number of points in the point cloud.

[0100] Specifically, without considering the translation matrix, the second optimal transformation formula is expressed as:

[0101]

[0102] Among them, P ti The coordinates of the point in the first point set that is closest to the second point set; P si represents the data points whose distance between the data points in the first point set and the data points in the second point set is less than the preset threshold; R represents the rotation matrix; n represents the number of data points whose distance between the data points in the first point set and the data points in the second point set is less than the preset threshold.

[0103] That can also be expressed as:

[0104] Equivalent to solving is the maximum, that is:

[0105]

[0106] Use SVD to decompose R*, trace(P t T RP s )=trace(∑V T RU) = trace(∑M), and finally get the optimal rotation matrix:

[0107]

[0108] S105, using a plane segmentation algorithm to segment the finely registered discrete point cloud data to obtain plane point cloud data and non-plane point cloud data.

[0109] S106, calculating the normal vector of each data point in the non-planar point cloud data within a preset range, and projecting the normal vector onto the Z axis in the three-dimensional coordinate axis to obtain a projection value, and determining the attributes of the data point according to the projection value; wherein the attributes include: plane points and elevation points. The elevation points are buildings, and the plane points are coal mines.

[0110] Exemplarily, the normal vector within a certain range of each point is calculated, and then the length unit of the normal vector is changed to obtain a normal vector with unit 1, and the normal vector with unit 1 is projected to the Z axis of the world coordinate system, and the plane and elevation points are determined based on the value projected to the Z axis.

[0111] In a specific embodiment provided by the present invention, the relationship between the projection value and 0.75 is determined, and points greater than 0.75 are considered planar points, and point cloud data with projection values ​​less than 0.75 are removed.

[0112] The non-planar point cloud data is divided according to the attributes of each data point, and the supervoxel clustering is used to obtain the clustering results. Figure 3 The following is the effect of the original marked building point cloud: Figure 4 The following is a visualization of the effect after supervoxel segmentation: Figure 5 Shown is the effect diagram after the coal mine segmentation of the mining area of ​​the present invention.

[0113] Specifically, the distance formula used in supervoxel clustering segmentation is expressed as:

[0114]

[0115] Among them, D c represents the Euclidean distance in normalized RGB space; D n Represents the normal vector distance; D s Indicates spatial distance; W c Indicates the weight of the color; W s represents the spatial distance weight; W n Represents the weight of the normal vector; R seed Indicates the spatial resolution.

[0116] The method for three-dimensional point cloud registration and fixed facility removal of coal mines in a warehouse environment of the present invention can obtain a large amount of discrete point cloud data containing coal mine information through high-precision laser radar scanning, providing a sufficient data basis for subsequent analysis. Reuse and reduce the amount of data to improve efficiency. Then, the point cloud is roughly registered according to the radar position information, and then the ICP algorithm is used for fine registration. Then, RANSAC is used to segment the ground, and the buildings and coal mines are distinguished according to the normal vectors, and then separated. Finally, supervoxel segmentation is used and a clustering algorithm is performed according to the convexity and concavity of the blocks to obtain the segmented coal mine point cloud data.

[0117] The invention provides a device for coal mine three-dimensional point cloud registration and fixed facility removal in a warehouse environment, the device comprising: a data acquisition module, a sampling module, a coarse fusion module, a fine fusion module, a data segmentation module and a clustering module.

[0118] The data acquisition module is used to use multiple high-precision laser radar devices to scan the coal mine area and obtain multiple discrete point cloud data sets.

[0119] The sampling module is used to preprocess multiple discrete point cloud data sets to obtain multiple sampled discrete point cloud data sets, and determine the first sampled discrete point cloud data set and the second sampled discrete point cloud data set corresponding to the laser radar device.

[0120] A coarse fusion module is used to calculate the first sampled discrete point cloud data set and the second sampled discrete point cloud data set according to the RANSAC registration algorithm and the ICP algorithm of the regional covariance feature to obtain a first translation matrix and a first rotation matrix, and perform coarse registration according to the first translation matrix and the first rotation matrix to obtain coarsely aligned discrete point cloud data, wherein the coarsely aligned discrete point cloud data includes: a first point set and a second point set.

[0121] The fine fusion module is used to calculate the coarsely aligned discrete point cloud data using the ICP algorithm to obtain the optimal translation matrix and the optimal rotation matrix, and use the optimal rotation matrix for fine alignment to obtain finely aligned discrete point cloud data.

[0122] The data segmentation module is used to segment the finely registered discrete point cloud data using a plane segmentation algorithm to obtain plane point cloud data and non-plane point cloud data.

[0123] The data partitioning module is used to calculate the normal vector of each data point in the non-planar point cloud data within a preset range, and project the normal vector onto the Z axis in the three-dimensional coordinate axis to obtain the projection value, and determine the attributes of the data point based on the projection value; wherein the attributes include: plane points and elevation points.

[0124] Although the present invention provides method operation steps such as embodiments or flow charts, more or fewer operation steps may be included based on conventional or non-creative labor. The order of steps listed in this embodiment is only one way of executing the order of many steps and does not represent the only execution order. When the device or client product in practice is executed, it can be executed in the order of the method shown in this embodiment or the accompanying drawings or in parallel (for example, in a parallel processor or multi-threaded processing environment).

[0125] The devices or modules described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above devices are described in various modules according to their functions. When implementing the present invention, the functions of each module can be implemented in the same or multiple software and / or hardware. Of course, the module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.

[0126] The method, device or module in the present invention can be implemented in the form of a computer-readable program code. The controller can be implemented in any appropriate manner. For example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (English: Application Specific Integrated Circuit; Abbreviation: ASIC), a programmable logic controller and an embedded microcontroller. Examples of controllers include but are not limited to the following microcontrollers: ARC 625D, AtmelAT91SAM, Microchip PIC18F26K20 and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code, the controller can be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller and an embedded microcontroller by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included in it for implementing various functions can also be regarded as structures within the hardware component. Or even, the means for realizing various functions may be regarded as both a software module for realizing the method and a structure within a hardware component.

[0127] Some modules in the apparatus of the present invention may be described in the general context of computer executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc. that perform specific tasks or implement specific abstract data types. The present invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0128] The present invention provides a server for three-dimensional point cloud registration and fixed facility removal of coal mines in a warehouse environment, the server comprising: a memory and a processor; the memory is used to store computer executable instructions; the processor is used to execute computer executable instructions to implement a method for three-dimensional point cloud registration and fixed facility removal of coal mines in a warehouse environment.

[0129] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present invention can be implemented by means of software plus necessary hardware. Based on such an understanding, the technical solution of the present invention can be essentially or partly contributed to the prior art in the form of a software product, or it can be reflected in the implementation process of data migration. The computer software product can be stored in a storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a mobile terminal, a server, or a network device, etc.) to execute the methods of various embodiments of the present invention or certain parts of the embodiments.

[0130] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. All or part of the present invention can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, mobile communication terminals, multi-processor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc.

[0131] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some or all of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the present invention.

Claims

1. A method for three-dimensional point cloud registration and fixed facility removal in a warehouse environment, characterized in that: include: Use multiple high-precision LiDAR devices to scan the coal mine area and obtain multiple discrete point cloud data sets; Preprocessing the multiple discrete point cloud data sets to obtain multiple sampled discrete point cloud data sets, and determining a first sampled discrete point cloud data set and a second sampled discrete point cloud data set corresponding to the laser radar device; According to the RANSAC registration algorithm and the ICP algorithm of the regional covariance feature, the first sampled discrete point cloud data set and the second sampled discrete point cloud data set are calculated to obtain a first translation matrix and a first rotation matrix, and a rough registration is performed according to the first translation matrix and the first rotation matrix to obtain rough aligned discrete point cloud data, wherein the rough aligned discrete point cloud data includes: a first point set and a second point set; Using the ICP algorithm to calculate the coarsely aligned discrete point cloud data to obtain an optimal translation matrix and an optimal rotation matrix, and using the optimal rotation matrix to perform fine alignment to obtain finely aligned discrete point cloud data; Segmenting the finely registered discrete point cloud data using a plane segmentation algorithm to obtain plane point cloud data and non-plane point cloud data; Calculate the normal vector of each data point in the non-planar point cloud data within a preset range, project the normal vector onto the Z axis in the three-dimensional coordinate axis to obtain a projection value, and determine the attributes of the data point based on the projection value; wherein the attributes include: plane points and elevation points.

2. The method for three-dimensional point cloud registration and fixed facility removal in a warehouse environment of coal mine according to claim 1, characterized in that: The preprocessing of the plurality of discrete point cloud data to obtain a first sampled discrete point cloud data set includes: Filtering the plurality of discrete point cloud data according to a first threshold value to obtain a filtered discrete point cloud data set; Downsampling is performed on the screened discrete point cloud data to obtain a first sampled discrete point cloud data set.

3. The method for three-dimensional point cloud registration and fixed facility removal in a warehouse environment of coal mine according to claim 1, characterized in that: The RANSAC registration algorithm and the ICP algorithm based on the regional covariance feature calculate the first sampled discrete point cloud data set and the second sampled discrete point cloud data set to obtain a first translation matrix and a first rotation matrix, including: Respectively calculating the covariance matrix corresponding to each discrete point cloud data in the first sampled discrete point cloud data set and the second sampled discrete point cloud data set; Calculate the distance between the covariance matrix corresponding to each discrete point cloud data in the first sampled discrete point cloud data set and the covariance matrix corresponding to each discrete point cloud data in the second sampled discrete point cloud data set according to the distance calculation formula; obtain the distance set corresponding to each discrete point cloud data in the first sampled discrete point cloud data set; Selecting discrete point cloud data in the second sampled discrete point cloud data set and discrete point cloud data in the first sampled discrete point cloud data set corresponding to the minimum distance in the distance set to form a matching point pair; A plurality of matching point pairs are determined, and a first translation matrix and a first rotation matrix are obtained by calculating the plurality of matching point pairs using an ICP algorithm.

4. The method for coal mine three-dimensional point cloud registration and fixed facility removal in a warehouse environment according to claim 3, characterized in that: The covariance matrix is ​​expressed as: in, is any point P in the first sampled discrete point cloud data set and the second sampled discrete point cloud data set i The three-dimensional centroid of the K nearest neighbor set, λ j represents the jth eigenvalue corresponding to the covariance matrix; It represents the jth eigenvector corresponding to the covariance matrix; k represents the number of adjacent points selected.

5. The method for coal mine three-dimensional point cloud registration and fixed facility removal in a warehouse environment according to claim 3, characterized in that: The distance calculation formula is expressed as: Where C1 represents a covariance matrix for one data point in the first sampled discrete point cloud data set; C2 represents a covariance matrix for one data point in the second sampled discrete point cloud data set; λ i represents the generalized eigenvalue; m represents the number of eigenvalues.

6. The method for coal mine three-dimensional point cloud registration and fixed facility removal in a warehouse environment according to claim 1, characterized in that: The method of calculating the roughly aligned discrete point cloud data using the ICP algorithm to obtain an optimal translation matrix and an optimal rotation matrix includes: Determine a first optimal transformation formula, perform a first-order derivative on the first optimal transformation formula, and set the derivative to 0 to obtain an optimal translation matrix; Without considering the translation error, a second optimal transformation formula is determined, and SVD decomposition calculation is performed on the second optimal transformation formula to obtain an optimal rotation matrix.

7. The method for three-dimensional point cloud registration and fixed facility removal in a warehouse environment of coal mine according to claim 6, characterized in that: The first optimal transformation formula is expressed as: Among them, P ti Indicates that the distance between the data point in the first point set and the data point in the second point set is less than the preset threshold; p si' represents the coordinates of each point in the second point set after the initial change is applied; R represents the rotation matrix; t represents the translation matrix; n represents the number of data points whose distance between the data points in the first point set and the data points in the second point set is less than a preset threshold.

8. The method for coal mine three-dimensional point cloud registration and fixed facility removal in a warehouse environment according to claim 6, characterized in that: The second optimal transformation formula is expressed as: Among them, P ti The coordinates of the point in the first point set that is closest to the second point set; P si represents the data points whose distance between the data points in the first point set and the data points in the second point set is less than the preset threshold; R represents the rotation matrix; n represents the number of data points whose distance between the data points in the first point set and the data points in the second point set is less than the preset threshold.

9. A coal mine three-dimensional point cloud registration and fixed facility removal device in a warehouse environment, characterized in that: include: The data acquisition module is used to use multiple high-precision laser radar devices to scan the coal mine area and obtain multiple discrete point cloud data sets; A sampling module, used to preprocess the multiple discrete point cloud data sets to obtain multiple sampled discrete point cloud data sets, and determine a first sampled discrete point cloud data set and a second sampled discrete point cloud data set corresponding to the laser radar device; A coarse fusion module is used to calculate the first sampled discrete point cloud data set and the second sampled discrete point cloud data set according to the RANSAC registration algorithm and the ICP algorithm of the regional covariance feature to obtain a first translation matrix and a first rotation matrix, and perform coarse registration according to the first translation matrix and the first rotation matrix to obtain coarsely aligned discrete point cloud data, wherein the coarsely aligned discrete point cloud data includes: a first point set and a second point set; A fine fusion module is used to calculate the coarsely aligned discrete point cloud data using an ICP algorithm to obtain an optimal translation matrix and an optimal rotation matrix, and to perform fine registration using the optimal rotation matrix to obtain finely registered discrete point cloud data; A data segmentation module, used to segment the finely registered discrete point cloud data using a plane segmentation algorithm to obtain plane point cloud data and non-plane point cloud data; A data partitioning module is used to calculate the normal vector of each data point in the non-planar point cloud data within a preset range, and project the normal vector onto the Z axis in the three-dimensional coordinate axis to obtain a projection value, and determine the attributes of the data point based on the projection value; wherein the attributes include: plane points and facade points.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium has executable instructions, and when the computer executes the executable instructions, it can implement the method for three-dimensional point cloud registration and fixed facility removal of coal mines in a warehouse environment as described in any one of claims 1-8.

Citation Information

Patent Citations

  • Point cloud registration method and device, electronic equipment and storage medium

    CN114219717A

  • Building indoor area prediction method and device, computer equipment and storage medium

    CN115147476A

  • Laser point cloud building contour progressive optimization method based on multi-dimensional features

    CN116824379A

  • Multi-platform laser point cloud registration method suitable for gingko man-made forest

    CN118172394A