An ICP matching method and system based on mine concave-convex characteristics

Abstract

The present invention relates to an ICP matching method and system based on the concave-convex characteristics of a mine tunnel. The present invention performs a first screening of a point cloud based on the first-class curvature and the first-class curvature threshold; performs a second screening of the point cloud based on the second-class curvature and the second-class curvature threshold of the remaining point cloud points; calculates the principal curvature of the remaining point cloud points after the two screenings, obtains the curvature features within the local neighborhood of the point cloud points, and screens feature points based on the local neighborhood curvature features; obtains the first-class curvature and principal curvature of the feature points in annular neighborhoods of different radii; selects the maximum absolute value of the first-class curvature and principal curvature in annular neighborhoods of different radii, and the annular neighborhood radius corresponding to the maximum value as the final feature; and in the process of executing the ICP algorithm, adds the final feature residual of the source and target point cloud frames as a constraint into the ICP objective function to perform point cloud registration. The concave-convex area features effectively alleviate the degradation problem of SLAM and improve the positioning and mapping accuracy in complex environments.

Description

Technical Field

[0001] The present invention relates to the technical field of positioning and mapping, and in particular to an ICP matching method based on the concave-convex characteristics of a mine tunnel. Background Art

[0002] With the continuous development of autonomous driving technology, simultaneous localization and mapping (SLAM) has gained increasing attention as a key technology. Currently, mainstream SLAM systems are mainly divided into laser SLAM and visual SLAM. Laser SLAM is widely used in autonomous driving due to its high accuracy and real-time performance, as it is not affected by illumination and directly acquires spatial position. However, in environments such as corridors and long, straight mine tunnels, the lack of forward direction constraints makes lidar prone to degradation, resulting in large SLAM mapping errors and affecting subsequent positioning and navigation. Zhang et al. analyzed the eigenvalues ​​of the constraint coefficient matrix of the problem and introduced a degradation factor to separate degenerate and non-degenerate directions, partially resolving the problem in directions with good conditions. Turcan et al. proposed the X-ICP algorithm, which groups eigenvectors by their projections on three axes of translation and rotation, determines whether each direction is localizable, and imposes corresponding constraints on non-localizable directions to mitigate degradation. These methods can alleviate the degradation problem to a certain extent, but in environments such as mines, the general direction of degradation is already clear, and there is a lack of research on the registration of similar point cloud features in the degradation direction. Summary of the Invention

[0003] In order to solve the above technical problems or at least partially solve the above technical problems, the present invention provides an ICP matching method and system based on the concave-convex characteristics of a mine.

[0004] In a first aspect, the present invention provides an ICP matching method based on the concave-convex characteristics of a mine, comprising:

[0005] Calibrate and align the LiDAR and inertial navigation unit, use the aligned LiDAR to collect the point cloud of the mine, pre-process the point cloud through filtering and voxelization, and construct an octree;

[0006] Screen out points in the flat area of ​​the point cloud according to the first-class curvature and the set first-class curvature threshold;

[0007] According to the second type curvature of the remaining point cloud points and the set second type curvature threshold, cylindrical structures, corridors and mine roof corners are screened out;

[0008] For the remaining point cloud points after two screenings, calculate their principal curvatures to obtain the final features of subsequent feature points for subsequent matching;

[0009] For the remaining point cloud after two screenings, the curvature features in the local neighborhood of the point cloud points are obtained, and the feature points are further screened based on the local neighborhood curvature features;

[0010] Set a set number of rings with different radii as the neighborhood of the feature point after screening, calculate the first kind of curvature and principal curvature of the feature point in the ring neighborhoods with different radii, and select the maximum absolute value of the first kind of curvature and principal curvature in the ring neighborhoods with different radii, as well as the ring neighborhood radius corresponding to the maximum value as the final feature of the feature point;

[0011] During the execution of the ICP algorithm, the final feature residuals corresponding to the feature points in the source and target point cloud frames are added as constraints into the ICP objective function to perform point cloud registration.

[0012] Furthermore, the process of screening out points in the flat area in the point cloud according to the first type curvature and the set first type curvature threshold includes:

[0013] The local neighborhood of each point in the point cloud is searched using an octree, and the covariance matrix is ​​calculated by analyzing the local neighborhood through principal component analysis. The eigenvalues ​​and eigenvectors of the covariance matrix are calculated, and the ratio of the smallest eigenvalue to the sum of all eigenvalues ​​is taken as the first-class curvature of the point. A first-class curvature threshold is set to compare with the first-class curvature of the point cloud point, and the point cloud points are screened, and the point cloud points with the first-class curvature higher than the first-class curvature threshold are retained.

[0014] Furthermore, the step of screening out cylindrical structures, corridors, and mine roof corners based on the second type curvature of the remaining point cloud points and a set second type curvature threshold includes:

[0015] The local neighborhood of each point in the point cloud is searched using an octree. The covariance matrix is ​​calculated by analyzing the local neighborhood using principal component analysis. The eigenvalues ​​and eigenvectors of the covariance matrix are calculated. The eigenvector corresponding to the minimum eigenvalue is used as the normal vector n of the point cloud point, and the direction of the normal vector is unified. A surface model is constructed and the local surface function is fitted using the point cloud coordinates in the local neighborhood by least squares fitting.

[0016] The first basic form coefficient and the second basic form coefficient are obtained by calculating the partial differential parameters according to the local surface function. The Gaussian curvature is calculated as the second kind of curvature using the first basic form coefficient and the second basic form coefficient:

[0017] The point cloud with the second type of curvature lower than the second type of curvature threshold is screened out by the preset second type of curvature threshold.

[0018] Furthermore, point cloud The local neighborhood of The quadratic surface model of is:

[0019] ;

[0020] in, and is the tangent plane projection coordinate, is the height along the normal direction, are the quadratic surface model parameters;

[0021] The least squares fitting uses the point cloud points in the local neighborhood Coordinates Solve for the parameters of the quadratic surface model, and the goal of the least squares fitting is to minimize the residual:

[0022] ;

[0023] Calculate partial differential parameters based on local surface functions:

[0024] Point cloud points based on fitting The local neighborhood of Local surface function Compute the first-order partial derivatives of a local surface:

[0025] ;

[0026] in, , ;

[0027] Point cloud points based on fitting The local neighborhood of Local surface function Compute the second-order partial derivatives of a local surface:

[0028] ;

[0029] Point cloud based The local neighborhood of The first order partial derivatives of the local surface function compute the first fundamental form coefficients:

[0030] ;

[0031] Point cloud point The local neighborhood of The second-order partial derivative of the local surface function is projected into the direction of the normal vector to obtain the second basic form coefficient:

[0032] ;

[0033] Calculate the point cloud points based on the first basic form coefficient and the second basic form coefficient Gaussian curvature of :

[0034] .

[0035] Furthermore, the process of calculating the principal curvature of a point cloud includes:

[0036] According to point cloud The local neighborhood of The first and second fundamental form coefficients of the local surface function calculate the mean curvature:

[0037] ;

[0038] Calculate its principal curvature based on Gaussian curvature and mean curvature :

[0039] .

[0040] Furthermore, for the point cloud remaining after the two screenings, obtaining the curvature features in the local neighborhood of the point cloud points, and further screening the feature points based on the local neighborhood curvature features includes:

[0041] For the remaining point cloud points after two screenings The local neighborhood of , extract the largest first and second curvatures in the local neighborhood as curvature features:

[0042] ,

[0043] ,

[0044] in, , Point cloud point The local neighborhood of Any point correspondence The first and second curvatures of the local neighborhood are selected as the point cloud points. The local neighborhood curvature characteristics of the point cloud point Delete points within the local neighborhood to ensure that the point cloud points There is only one set of local neighborhood curvature features within the local neighborhood.

[0045] Furthermore, the final feature residual constraint is added to the ICP objective function:

[0046] ;

[0047] in, For IPC point pair matching, is the transformation relationship, including rotation and translation transformation, is the feature point i of the source point cloud frame s, is the matched feature point i of the target point cloud frame t; is the final feature residual constraint term, Represents the global weight factor, balancing the contribution of the IPC point to the matching term and the final feature residual constraint term, Represents the final feature residual of the feature point indexed by i between the source point cloud frame s and the target point cloud frame t.

[0048] Furthermore, the final feature residual is:

[0049] ;

[0050] in, Represents the final feature residual of the feature point indexed by i between the source point cloud frame s and the target point cloud frame t, Indicates the weight of the lth feature element in the corresponding final feature, Point represents the lth feature element of the final feature of the feature point indexed as i in the source point cloud frame s and the target point cloud frame t.

[0051] Furthermore, after introducing the final feature residual constraint term, the final total Jacobian matrix of the transformation relation T in the ICP algorithm is:

[0052] ;

[0053] in, , is the Jacobian matrix of the IPC point pair matching term, , is the Jacobian matrix of the final characteristic residual constraint term.

[0054] In second aspect, the present invention provides an ICP matching system based on the concave-convex characteristics of mine tunnels, comprising: at least one processing unit, the processing unit being connected to a storage unit and an acquisition unit via a bus unit, the storage unit storing a computer program and data collected by the acquisition unit, the processing unit implementing the ICP matching method based on the concave-convex characteristics of mine tunnels by running the computer program stored in the storage unit.

[0055] The above technical solution provided by the embodiment of the present invention has the following advantages compared with the prior art:

[0056] This application proposes an ICP matching method based on the concave and convex characteristics of mine tunnels. It performs point cloud screening based on two types of curvature, and uses a multi-scale annular neighborhood method to achieve targeted feature extraction of the concave and convex areas of the mine tunnel to obtain the final features. The final feature residual is introduced into the final ICP objective function. The concave and convex area features strengthen the constraint ability of SLAM in the degradation direction, which can effectively alleviate the degradation problem of SLAM and improve the positioning and mapping accuracy in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0059] Figure 1 A flowchart of an ICP matching method based on the concave-convex characteristics of a mine provided by an embodiment of the present invention;

[0060] Figure 2 A flowchart of filtering out points in a flat area in a point cloud according to a first type of curvature and a set first type of curvature threshold provided in an embodiment of the present invention;

[0061] Figure 3 A flowchart for screening out cylindrical structures, corridors, and mine roof corners based on the second type curvature of the remaining point cloud points and a set second type curvature threshold provided in an embodiment of the present invention;

[0062] Figure 4 Schematic diagram of an ICP matching system based on the concave-convex characteristics of a mine provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described 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 making creative efforts shall fall within the scope of protection of the present invention.

[0064] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0065] Example 1

[0066] See Figure 1 As shown, the ICP matching method based on the concave-convex characteristics of the mine provided in this application includes:

[0067] Step 1: Calibrate and align the LiDAR and IMU to ensure that the point cloud data is synchronized with the attitude information of the IMU and ensure the accuracy of subsequent registration.

[0068] The aligned lidar is used to collect the point cloud of the mine, and the point cloud is preprocessed by filtering and voxelization to construct an octree.

[0069] Step 2: Screen out points in the flat area of ​​the point cloud based on the first-class curvature and the set first-class curvature threshold.

[0070] like Figure 2 As shown, in step 2, the local neighborhood of each point in the point cloud is searched using an octree, and the covariance matrix is ​​calculated by principal component analysis. The eigenvalues ​​and eigenvectors of the covariance matrix are calculated, and the ratio of the smallest eigenvalue to the sum of all eigenvalues ​​is used as the first-class curvature of the point. Set a first-class curvature threshold The first-class curvature of the point cloud is compared with the first-class curvature of the point cloud points to filter the point cloud points and retain the point cloud points whose first-class curvature is higher than the first-class curvature threshold. The retained point cloud points are edge or corner features.

[0071] In the specific implementation process, the formula for calculating the covariance matrix of each point using the neighboring points in the local neighborhood of the point is as follows:

[0072] ;

[0073] in, Point cloud point The local neighborhood of The center of mass of the point cloud, the coordinates of the center of mass are in the local neighborhood The average of the coordinates of the midpoints, Point cloud point The covariance matrix of

[0074] The formula for calculating the eigenvalues ​​and eigenvectors of the point cloud point covariance matrix is ​​as follows:

[0075] ;

[0076] in, Represents point cloud points The covariance matrix of The eigenvalues ​​and eigenvectors of the covariance matrix are obtained by eigendecomposition of the covariance matrix to obtain three eigenvalues ​​and corresponding eigenvectors.

[0077] The ratio of the smallest eigenvalue to the sum of all eigenvalues ​​is taken as the first kind curvature of the point:

[0078]

[0079] in, Point cloud point The first kind of curvature. The relationship between the eigenvalues ​​is: , because the size of the eigenvalue is related to the direction of the point distribution within the local neighborhood. For example, if the point is on a plane, the smallest eigenvalue will be small, while the two larger eigenvalues ​​will be large, so the curvature will be small. Conversely, if the point is at an edge or corner, the curvature will be high.

[0080] Step 3: Screen out cylindrical structures, corridors, and mine roof corners based on the second-category curvature of the remaining point cloud points and the set second-category curvature threshold.

[0081] like Figure 3 As shown in FIG, the process of step three includes: using the octree to search the local neighborhood of each point in the point cloud, analyzing the local neighborhood through principal component analysis to calculate the covariance matrix, calculating the eigenvalues ​​and eigenvectors of the covariance matrix, and using the eigenvector corresponding to the minimum eigenvalue as the normal vector n of the point cloud point, unifying the direction of the normal vector; using the normal vector to fit the local surface function to the remaining point cloud, and using the fitted local surface function to calculate the second kind of curvature.

[0082] During the implementation, an octree is used to search the local neighborhood of each point in the point cloud. Principal component analysis is used to analyze the local neighborhood and calculate the covariance matrix. The eigenvalues ​​and eigenvectors of the covariance matrix are calculated, as already performed in step 2. The eigenvector corresponding to the minimum eigenvalue in step 2 is directly used as the normal vector n of the point cloud point, unifying the normal vector direction. A surface model is constructed, and the local surface function is fitted using the point cloud coordinates within the local neighborhood through least squares fitting. In this example, a quadratic surface model is used to project the local neighborhood point set onto the tangent plane defined by the normal vector n, and the quadratic surface model is used to fit the local surface structure.

[0083] Point Cloud Points The local neighborhood of The quadratic surface model of is:

[0084] ;

[0085] in, and is the tangent plane projection coordinate, is the height along the normal direction, are the quadratic surface model parameters.

[0086] The least squares fitting uses the point cloud points in the local neighborhood Coordinates Solve for the parameters of the quadratic surface model. The goal of the least squares fit is to minimize the residual:

[0087] ;

[0088] Calculate partial differential parameters based on local surface functions:

[0089] Point cloud points based on fitting The local neighborhood of Local surface function Compute the first-order partial derivatives of a local surface:

[0090] ;

[0091] in, , .

[0092] Point cloud points based on fitting The local neighborhood of Local surface function Compute the second-order partial derivatives of a local surface:

[0093] ;

[0094] Point cloud based The local neighborhood of The first order partial derivatives of the local surface function compute the first fundamental form coefficients:

[0095] ;

[0096] Point cloud point The local neighborhood of The second-order partial derivative of the local surface function is projected into the direction of the normal vector to obtain the second basic form coefficient:

[0097] .

[0098] Calculate the point cloud points based on the first basic form coefficient and the second basic form coefficient Gaussian curvature of :

[0099] ;

[0100] Gaussian curvature as curvature of the second kind;

[0101] Calculate the Gaussian curvature of each remaining point cloud point, and use the preset second-class curvature threshold to filter out the point cloud with a second-class curvature lower than the second-class curvature threshold.

[0102] Step 4: For the remaining point cloud points after two screenings, calculate their principal curvatures as the final features of the points for subsequent matching.

[0103] The process of calculating the principal curvature of a point cloud point includes:

[0104] According to point cloud The local neighborhood of The first and second fundamental form coefficients of the local surface function calculate the mean curvature:

[0105] ;

[0106] Calculate its principal curvature based on Gaussian curvature and mean curvature :

[0107] .

[0108] Step 5: For the remaining point cloud after two screenings, obtain the curvature features in the local neighborhood of the point cloud points, and further filter the feature points based on the local neighborhood curvature features.

[0109] For the remaining point cloud points after two screenings The local neighborhood of , extract the largest first and second curvatures in the local neighborhood as curvature features:

[0110] ,

[0111] ,

[0112] in, , Point cloud point The local neighborhood of Any point correspondence The first and second curvatures of the local neighborhood are selected as the point cloud points. The local neighborhood curvature characteristics of the point cloud point Delete points within the local neighborhood to ensure that the point cloud points There is only one set of local neighborhood curvature features within the local neighborhood.

[0113] Step 6: Set a set number of rings of different radii as the neighborhood of the filtered feature points, calculate the first kind of curvature and principal curvature of the feature points in the ring neighborhoods of different radii, and select the maximum absolute value of the first kind of curvature and principal curvature in the ring neighborhoods of different radii, as well as the ring neighborhood radius corresponding to the maximum value as the final feature of the feature point:

[0114] ;

[0115] in, is the maximum value of the first kind of curvature of the feature points in each annular neighborhood; is the principal curvature of the feature points in each annular neighborhood The maximum value of ; where is the set of feature points in the annular neighborhood; is the principal curvature of the feature points in each annular neighborhood The absolute maximum value of . is the radius of the annular neighborhood corresponding to the maximum value.

[0116] Step 7: During the execution of the ICP algorithm, the final feature residuals corresponding to the feature points in the source and target point cloud frames are added as constraints into the ICP objective function to perform point cloud registration.

[0117] Calculate the final feature residual:

[0118] ;

[0119] in, Represents the final feature residual of the feature point indexed by i between the source point cloud frame s and the target point cloud frame t, Indicates the weight of the lth feature element in the corresponding final feature, Point represents the lth feature element of the final feature of the feature point indexed as i in the source point cloud frame s and the target point cloud frame t.

[0120] Add the final feature residual constraint to the ICP objective function:

[0121] ;

[0122] in, For IPC point pair matching, is the transformation relationship, including rotation and translation transformation, is the feature point i of the source point cloud frame s, is the matched feature point i of the target point cloud frame t; is the final feature residual constraint term, Represents the global weight factor, balancing the contribution of the IPC point to the matching term and the final feature residual constraint term.

[0123] The final total Jacobian matrix of the corresponding transformation relation T is:

[0124] ;

[0125] in, , is the Jacobian matrix of the IPC point pair matching term,

[0126] , is the Jacobian matrix of the final characteristic residual constraint term.

[0127] The Gauss-Newton method is used to minimize the objective function, and T is updated iteratively until the objective function converges.

[0128] Example 2

[0129] See Figure 4 As shown, an embodiment of the present invention provides an ICP matching system based on the concave-convex characteristics of a mine tunnel, comprising: at least one processing unit, the processing unit connected to a storage unit and an acquisition unit via a bus unit, the acquisition unit comprising a laser radar and an inertial navigation unit, the storage unit as a computer-readable storage medium, capable of storing software programs, computer executable programs, and modules, such as the software programs, computer executable programs, and modules corresponding to an ICP matching method based on the concave-convex characteristics of a mine tunnel in an embodiment of the present invention, which can be used to store data collected by the acquisition unit. The processing unit implements the above-mentioned ICP matching method based on the concave-convex characteristics of a mine tunnel by running the software programs, computer executable programs, and modules stored in the storage unit, comprising:

[0130] Calibrate and align the LiDAR and inertial navigation unit, use the aligned LiDAR to collect the point cloud of the mine, pre-process the point cloud through filtering and voxelization, and construct an octree;

[0131] Screen out points in the flat area of ​​the point cloud according to the first-class curvature and the set first-class curvature threshold;

[0132] According to the second type curvature of the remaining point cloud points and the set second type curvature threshold, cylindrical structures, corridors and mine roof corners are screened out;

[0133] For the remaining point cloud points after two screenings, calculate their principal curvatures to obtain the final features of subsequent feature points for subsequent matching;

[0134] For the remaining point cloud after two screenings, the curvature features in the local neighborhood of the point cloud points are obtained, and the feature points are further screened based on the local neighborhood curvature features;

[0135] Set a set number of rings with different radii as the neighborhood of the feature point after screening, calculate the first kind of curvature and principal curvature of the feature point in the ring neighborhoods with different radii, and select the maximum absolute value of the first kind of curvature and principal curvature in the ring neighborhoods with different radii, as well as the ring neighborhood radius corresponding to the maximum value as the final feature of the feature point;

[0136] During the execution of the ICP algorithm, the final feature residuals corresponding to the feature points in the source and target point cloud frames are added as constraints into the ICP objective function to perform point cloud registration.

[0137] Of course, the computer program stored in the storage unit of the ICP matching system based on the concave-convex characteristics of a mine tunnel provided by an embodiment of the present invention is not limited to the method operations described above, and can also execute related operations in the ICP matching method based on the concave-convex characteristics of a mine tunnel provided by any embodiment of the present invention.

[0138] Example 3

[0139] An embodiment of the present invention provides a computer-readable storage medium storing a computer program. When the computer program is executed, the ICP matching method based on the concave-convex characteristics of a mine tunnel is implemented, including:

[0140] Calibrate and align the LiDAR and inertial navigation unit, use the aligned LiDAR to collect the point cloud of the mine, pre-process the point cloud through filtering and voxelization, and construct an octree;

[0141] Screen out points in the flat area of ​​the point cloud according to the first-class curvature and the set first-class curvature threshold;

[0142] According to the second type curvature of the remaining point cloud points and the set second type curvature threshold, cylindrical structures, corridors and mine roof corners are screened out;

[0143] For the remaining point cloud points after two screenings, calculate their principal curvatures to obtain the final features of subsequent feature points for subsequent matching;

[0144] For the remaining point cloud after two screenings, the curvature features in the local neighborhood of the point cloud points are obtained, and the feature points are further screened based on the local neighborhood curvature features;

[0145] Set a set number of rings with different radii as the neighborhood of the feature point after screening, calculate the first kind of curvature and principal curvature of the feature point in the ring neighborhoods with different radii, and select the maximum absolute value of the first kind of curvature and principal curvature in the ring neighborhoods with different radii, as well as the ring neighborhood radius corresponding to the maximum value as the final feature of the feature point;

[0146] During the execution of the ICP algorithm, the final feature residuals corresponding to the feature points in the source and target point cloud frames are added as constraints into the ICP objective function to perform point cloud registration.

[0147] An embodiment of the present invention provides a computer-readable storage medium, the computer program stored in which is not limited to the method operations described above, but can also execute related operations in an ICP matching method based on mine concave-convex characteristics provided by any embodiment of the present invention.

[0148] In the embodiments provided by the present invention, it should be understood that the disclosed structures and methods can be implemented in other ways. For example, the structural embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, structure or unit, which can be electrical, mechanical or other forms.

[0149] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0150] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0151] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is intended to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. An ICP matching method based on the concave-convex characteristics of a mine, characterized by: include: Calibrate and align the LiDAR and inertial navigation unit, use the aligned LiDAR to collect the point cloud of the mine, pre-process the point cloud through filtering and voxelization, and construct an octree; Screen out points in the flat area of ​​the point cloud according to the first-class curvature and the set first-class curvature threshold; According to the second type curvature of the remaining point cloud points and the set second type curvature threshold, cylindrical structures, corridors and mine roof corners are screened out; For the remaining point cloud points after two screenings, calculate their principal curvatures to obtain the final features of subsequent feature points for subsequent matching; For the remaining point cloud after two screenings, the curvature features in the local neighborhood of the point cloud points are obtained, and the feature points are further screened based on the local neighborhood curvature features; Set a set number of rings with different radii as the neighborhood of the feature point after screening, obtain the first kind of curvature and principal curvature of the feature point in the ring neighborhoods with different radii, and select the maximum absolute value of the first kind of curvature and principal curvature in the ring neighborhoods with different radii, as well as the ring neighborhood radius corresponding to the maximum value as the final feature of the feature point; During the execution of the ICP algorithm, the final feature residuals corresponding to the feature points in the source and target point cloud frames are added as constraints into the ICP objective function to perform point cloud registration.

2. The ICP matching method based on the concave-convex characteristics of a mine according to claim 1, characterized in that: The process of screening out points in a flat area in the point cloud according to the first type curvature and a set first type curvature threshold comprises: The local neighborhood of each point in the point cloud is searched using an octree, and the covariance matrix is ​​calculated by analyzing the local neighborhood through principal component analysis. The eigenvalues ​​and eigenvectors of the covariance matrix are calculated, and the ratio of the smallest eigenvalue to the sum of all eigenvalues ​​is taken as the first-class curvature of the point. A first-class curvature threshold is set to compare with the first-class curvature of the point cloud point, and the point cloud points are screened, and the point cloud points with the first-class curvature higher than the first-class curvature threshold are retained.

3. The ICP matching method based on the concave-convex characteristics of a mine according to claim 1, characterized in that: The method of screening out cylindrical structures, corridors, and corners of mine roofs according to the second type curvature of the remaining point cloud points and the set second type curvature threshold includes: The local neighborhood of each point in the point cloud is searched using an octree. The covariance matrix is ​​calculated by analyzing the local neighborhood using principal component analysis. The eigenvalues ​​and eigenvectors of the covariance matrix are calculated. The eigenvector corresponding to the minimum eigenvalue is used as the normal vector n of the point cloud point, and the direction of the normal vector is unified. A surface model is constructed and the local surface function is fitted using the point cloud coordinates in the local neighborhood by least squares fitting. The first basic form coefficient and the second basic form coefficient are obtained by calculating the partial differential parameters according to the local surface function. The Gaussian curvature is calculated as the second kind of curvature using the first basic form coefficient and the second basic form coefficient: The point cloud with the second type of curvature lower than the second type of curvature threshold is screened out by the preset second type of curvature threshold.

4. The ICP matching method based on the concave-convex characteristics of a mine according to claim 3, characterized in that: Point Cloud Points The local neighborhood of The quadratic surface model of is: ； in, and is the tangent plane projection coordinate, is the height along the normal direction, are the quadratic surface model parameters; The least squares fitting uses the point cloud points in the local neighborhood Coordinates Solve for the parameters of the quadratic surface model, and the goal of the least squares fitting is to minimize the residual: ； Calculate partial differential parameters based on local surface functions: Point cloud points based on fitting The local neighborhood of Local surface function Compute the first-order partial derivatives of a local surface: ； in, , ; Point cloud points based on fitting The local neighborhood of Local surface function Compute the second-order partial derivatives of a local surface: ； Point cloud based The local neighborhood of The first order partial derivatives of the local surface function compute the first fundamental form coefficients: ； Point cloud point The local neighborhood of The second-order partial derivative of the local surface function is projected into the direction of the normal vector to obtain the second basic form coefficient: ； Calculate the point cloud points based on the first basic form coefficient and the second basic form coefficient Gaussian curvature of : 。 5. The ICP matching method based on the concave-convex characteristics of a mine according to claim 4, characterized in that: The process of calculating the principal curvature of a point cloud point includes: According to point cloud The local neighborhood of The first and second fundamental form coefficients of the local surface function calculate the mean curvature: ； Calculate its principal curvature based on Gaussian curvature and mean curvature : 。 6. The ICP matching method based on the concave-convex characteristics of a mine according to claim 1, characterized in that: The method of obtaining curvature features in a local neighborhood of the point cloud points remaining after two screenings and further screening feature points based on the curvature features in the local neighborhood includes: For the remaining point cloud points after two screenings The local neighborhood of , extract the largest first and second curvatures in the local neighborhood as curvature features: ， ， in, , Point cloud point The local neighborhood of Any point correspondence The first and second curvatures of the local neighborhood are selected as the point cloud points. The local neighborhood curvature characteristics of the point cloud point Delete points within the local neighborhood to ensure that the point cloud points There is only one set of local neighborhood curvature features within the local neighborhood.

7. The ICP matching method based on the concave-convex characteristics of a mine according to claim 1, characterized in that: Add the final feature residual constraint to the ICP objective function: ； in, For IPC point pair matching, is the transformation relationship, including rotation and translation transformation, is the feature point i of the source point cloud frame s, is the matched feature point i of the target point cloud frame t; is the final feature residual constraint term, Represents the global weight factor, balancing the contribution of the IPC point to the matching term and the final feature residual constraint term, Represents the final feature residual of the feature point indexed by i between the source point cloud frame s and the target point cloud frame t.

8. The ICP matching method based on the concave-convex characteristics of a mine according to claim 7, characterized in that: Final feature residual: ； in, Represents the final feature residual of the feature point indexed by i between the source point cloud frame s and the target point cloud frame t, Indicates the weight of the lth feature element in the corresponding final feature, Point represents the lth feature element of the final feature of the feature point indexed as i in the source point cloud frame s and the target point cloud frame t.

9. The ICP matching method based on the concave-convex characteristics of a mine according to claim 7, characterized in that: After introducing the final feature residual constraint term, the final total Jacobian matrix of the transformation relation T in the ICP algorithm is: ； in, , is the Jacobian matrix of the IPC point pair matching term, , is the Jacobian matrix of the final characteristic residual constraint term.

10. An ICP matching system based on the concave-convex characteristics of a mine, comprising: At least one processing unit, wherein the processing unit is connected to the storage unit and the acquisition unit via a bus unit, wherein the storage unit stores a computer program and data collected by the acquisition unit, and the processing unit implements the ICP matching method based on the concave-convex characteristics of the mine as described in any one of claims 1 to 9 by running the computer program stored in the storage unit.

Images