A calculation method for the center line of a complex roadway network based on skeleton contraction

Abstract

The present invention discloses a method for calculating the centerline of a complex roadway network based on skeleton contraction, belonging to the technical field of three-dimensional modeling and point cloud data processing of underground mines, and is used for calculating the centerline of a complex roadway network. It includes establishing a KDTree for the underground mine roadway point cloud data to obtain the neighborhood set of points, calculating the geometric features and neighborhood distances of points according to the neighborhood point set; performing iterative contraction on the underground mine roadway point cloud data; refining the skeleton according to the geometric features of each contracted skeleton point, and removing discrete noise points by combining with the octree filtering method; processing the point set in the intersection area by using the principal component method and the DBScan algorithm, judging the intersection type, and performing refined calculation of the center point for each intersection; fitting the disordered center points into an ordered centerline. The present invention can accurately calculate the overall roadway centerline of the underground mine roadway project, overcome the complex structural features of intersections and staggered floors, and achieve refined calculation for different roadway intersection types.

Description

Technical Field

[0001] The present invention discloses a method for calculating the center line of a complex roadway network based on skeleton contraction, belonging to the technical field of three-dimensional modeling and point cloud data processing in underground mines. Background Technique

[0002] Underground mine roadways are an important foundation and main component of mine resource exploitation. The accurate grasp of their spatial geometric information is the key to realizing digital and intelligent management of mines. As the skeleton structure of the roadway spatial geometry, the center line of the roadway is not only an important data basis for roadway modeling, navigation, monitoring and maintenance, but also directly related to the production efficiency and safety management level of mines. Due to the complex structure of the roadway network in underground mine roadway engineering, its diversity and intersection bring great difficulties and challenges to the acquisition of the center line. In recent years, with the development of three-dimensional laser scanning technology and point cloud data processing technology, the acquisition of high-precision point cloud data has become more convenient, providing strong support for the calculation of the roadway center line. Therefore, realizing the efficient calculation of the center line of the complex roadway network in underground mine roadway engineering based on three-dimensional laser point cloud data is an important means to improve the mine management level.

[0003] The present invention proposes a method for calculating the center line of a complex roadway network based on skeleton contraction, adopting a calculation process from rough to fine. Based on the geometric characteristics of roadway point clouds and filtering methods, skeleton rough contraction and refinement are carried out. According to the morphological characteristics of Y-shaped and T-shaped intersections in the roadway, different processing methods are designed respectively to realize the accurate calculation of the center line, and it can solve the problem that it is difficult to calculate the center line at complex positions such as roadway intersections and staggered floors, providing effective spatial geometric information for realizing digital and intelligent management of mines. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for calculating the center line of a complex roadway network based on skeleton contraction to solve the problem that it is difficult to calculate the center line of a roadway network with a complex structure in underground mine roadway engineering in the prior art.

[0005] A method for calculating the center line of a complex roadway network based on skeleton contraction includes:

[0006] S1. For the underground mine roadway point cloud data Build a KDTree to obtain the neighborhood distance for initial contraction, read the point cloud data point by point and calculate the neighborhood point set of each point, and calculate the surface feature and line feature of each point from the eigenvalue of the neighborhood point set;

[0007] S2. Perform iterative contraction on the underground mine roadway point cloud data, continuously update the coordinate value, plane feature, linear feature and neighborhood range of each point, and obtain the skeleton points of rough contraction;

[0008] S3. Refine the skeleton according to the geometric features of each shrunk skeleton point, and combine with the octree filtering method to remove discrete noise points, obtaining the central points of straight and curved roadways and the point set of the intersection area;

[0009] S4. By analyzing the morphological characteristics of underground mine roadways, define Y-shaped and T-shaped roadway intersections, process the point set of the intersection area using the principal component method and the DBScan algorithm, judge the intersection type, and perform refined calculation of the central point for each intersection;

[0010] S5. Smoothly connect the refined central points with other central points, and use the minimum spanning tree to organize the disordered central points, fitting the disordered central points into an ordered centerline.

[0011] S1 includes, S1.1. For the th point cloud data , , being the total number of point cloud data, set the distance neighborhood D for neighborhood search according to the distance to the second nearest point. D is:

[0012] ;

[0013] Obtain the neighborhood point set with D as the radius , , , and being the th neighborhood point's , and values, being the total number of neighborhoods.

[0014] S1 includes, S1.2. Calculate the distance mean of the neighborhood range based on the neighborhood point set:

[0015] ;

[0016] In the formula, , and are respectively the ranges of in the , and directions, and are respectively the maximum and minimum values of in the , and directions; ​

[0017] Using as the neighborhood radius, re-search the neighborhood point set, and calculate the eigenvalues according to the covariance matrix of the neighborhood point set 、 and to obtain the line feature and the surface feature :

[0018] .

[0019] S2 includes, S2.1. After the initial shrinkage of the skeleton based on the neighborhood, the point cloud data is updated to points with 6 attribute values ;

[0020] S2.2. Perform iterative shrinkage. If is close to 0 and is close to 1, it means that the current point presents a surface feature rather than a skeleton line. Update the 、 、 and values of this point with the coordinate mean of the neighborhood point set re-searched with the current point , 、 and are the 、 、 and values of the -th neighborhood point re-searched respectively, and

[0021] is the total number of neighborhood points re-searched:

[0022] In the formula, 、 and are the updated 、 and values respectively;

[0023] Update the neighborhood distance:

[0024] ;

[0025] In the formula, is the distance mean of the updated neighborhood range;

[0026] Continuously update , and the shrinkage round is 4.

[0027] S3 includes, S3.1. Analyze the skeleton points after rough shrinkage, and further refine the rough skeleton line by building a KDTree

[0028] S3.2. Re-update of 、 and values, according to the value of the point cloud data, find the neighborhood point set within twice the range, and find the point in the point set that is farthest from ;

[0029] S3.3. If the value of is greater than 0.2, it means that is not a point on the skeleton line and needs to be refined. Calculate the mean value of the and values of 、 and to update the coordinate value of , and update the value of to the distance to the nearest point to ;

[0030] S3.4. Perform octree filtering on the refined data again to remove discrete points and obtain the roughly calculated centerline.

[0031] S4 includes S4.1. Perform principal component analysis on the neighborhood of each point in the roughly calculated centerline, and calculate the point set of the intersection area where the difference between the first principal component and the second principal component is less than the threshold 13;

[0032] S4.2. Use the DBScan algorithm to cluster the point set of the intersection area, and each intersection is clustered into one class.

[0033] S4 includes S4.3. According to the structural characteristics of the mine roadway, define the roadway intersections as Y-shaped and T-shaped. The point set of the intersection area presents a triangular feature, and obtain three vertices 、 and ;

[0034] Connect the three obtained vertices into the three sides of a triangle, and obtain the respective center points lmean_01, lmean_12, and lmean_20 of the three sides;

[0035] Extend the vertices respectively to the centerlines they are connected to, and obtain the respective distal points max_distance0, max_distance1, and max_distance2 of the vertices;

[0036] Connect the distal points to the central point to obtain three angles angle_01, angle_12, and angle_02 that can reflect the intersection type;

[0037] Calculate the difference between the smallest angle and the middle-sized angle. If the difference is less than 10°, it is a T-shaped intersection; otherwise, it is a Y-shaped intersection.

[0038] S4 includes S4.4. For a Y-shaped intersection, the side corresponding to the largest angle is the centerline of the straight lane. Directly connect the two distal points corresponding to the largest angle to generate a straight line, and evenly generate straight-line points according to the point cloud density as the points of the centerline. The side corresponding to the middle angle is the bifurcated lane of the large lane, showing the characteristic of curving outwards. Use the cubic B-spline curve fitting method to calculate the center point of the lane. The side corresponding to the smallest angle is not the centerline, and directly delete the side corresponding to the smallest angle.

[0039] S4 includes S4.5. For a T-shaped intersection, the side corresponding to the largest angle is the centerline of the straight lane. Directly connect the two distal points corresponding to the largest angle to generate a straight line, and evenly generate straight-line points according to the point cloud density as the points of the centerline. And generate a new skeleton line between the third point and this center point. After processing different intersection area point sets respectively, the accurately calculated centerline of the lane is obtained.

[0040] S5 includes S5.1. Mark and sort the obtained center points, store the generated center points in an array and create a new array and initialize it to -1, indicating that the corresponding center point has not been sorted yet, indicating that the corresponding center point enters the sorting sequence, indicating the th element in the array;

[0041] S5.2. Use the minimum spanning tree to connect the center points of the straight lanes and the center points of the intersecting lanes respectively. Each section forms the shortest connected line. Finally, connect the endpoints of the accurately extracted centerline of the intersecting lanes with the centerline of the straight lanes to generate the complete centerline of the lane.

[0042] Compared with the prior art, the present invention has the following beneficial effects: The present invention can calculate the centerline of the underground mine roadway engineering with complex characteristics such as cross and staggered layers; avoid the problem of overlap of roadways at different height layers through skeleton contraction and refinement; at the same time, define different roadway intersection types, calculate the centerline more precisely, and better reflect the spatial geometric characteristics of the roadway engineering, providing an effective geometric feature data basis for realizing the digital and intelligent management of the mine. Brief Description of the Drawings

[0043] Figure 1 is the technical flow chart of the present invention;

[0044] Figure 2 is the calculation result chart of the center line thickness;

[0045] Figure 3 is the calculation and clustering chart of the point set in the intersection area;

[0046] Figure 4 is the schematic diagram of a Y-shaped intersection;

[0047] Figure 5 is the schematic diagram of a T-shaped intersection;

[0048] Figure 6 is the refined calculation result chart of the Y-shaped intersection;

[0049] Figure 7 is the refined calculation result chart of the T-shaped intersection;

[0050] Figure 8 is the calculation result chart of the center line. Specific implementation manner

[0051] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0052] A method for calculating the center line of a complex roadway network based on skeleton contraction, as Figure 1 includes:

[0053] S1. Establish a KDTree for the underground mine roadway point cloud data to perform initial contraction to obtain the neighborhood distance, read the point cloud data point by point and calculate the neighborhood point set of each point, and calculate the surface feature and line feature of each point from the eigenvalue of the neighborhood point set;

[0054] S2. Perform iterative contraction on the underground mine roadway point cloud data, continuously update the coordinate value, plane feature, linear feature and neighborhood range of each point, and obtain the roughly contracted skeleton points as Figure 2 shown;

[0055] S3. Refine the skeleton according to the geometric features of each contracted skeleton point, and combine the octree filtering method to remove discrete noise points to obtain the center points of straight roadways and curved roadways and the point set in the intersection area;

[0056] S4. By analyzing the morphological characteristics of underground mine roadways, define Y-shaped and T-shaped roadway intersections, process the point set in the intersection area using the principal component method and the DBScan algorithm as Figure 3 shown, and determine the intersection type, and perform refined calculation of the center point for each intersection;

[0057] S5. Smoothly connect the refined center points with other center points, and use the minimum spanning tree to organize the disordered center points, and fit the disordered center points into an ordered center line.

[0058] S1 includes, S1.1. For the th point cloud data , , being the total number of point cloud data, set the distance neighborhood D for neighborhood search according to the distance to the second nearest point. D is: ;

[0059] ;

[0060] Obtain the neighborhood point set with D as the radius , , , and being the th neighborhood point's , and values, being the total number of neighborhoods.

[0061] S1 includes, S1.2. Calculate the average distance within the neighborhood range based on the neighborhood point set:

[0062] ;

[0063] In the formula, , and are respectively the ranges of in the , and directions, and are respectively the maximum and minimum values of in the , and directions;

[0064] Re-search the neighborhood point set with as the neighborhood radius, and calculate the eigenvalues , and to obtain line features and surface features :

[0065] .

[0066] S2 includes. S2.1. After the initial shrinkage of the skeleton based on the neighborhood, the point cloud data is updated to points with 6 attribute values ;

[0067] S2.2. Perform iterative shrinkage. If is close to 0, is close to 1, it means that the current point presents a surface feature rather than a skeleton line. Update the of this point with the coordinate mean of the neighborhood point set , and values, , , and are the th , and values of the re-searched neighborhood points respectively, is the total number of re-searched neighborhood points:

[0068] ;

[0069] In the formula, , and are the updated , and values respectively;

[0070] Update the neighborhood distance:

[0071] ;

[0072] In the formula, is the distance mean of the updated neighborhood range;

[0073] Continuously update , and the shrinkage round is 4.

[0074] S3 includes. S3.1. Analyze the skeleton points after rough shrinkage, and further refine the rough skeleton line by building a KDTree;

[0075] S3.2. Re-update the of , and Value, according to the point cloud data Value, find twice The range of the neighborhood point set, find the point in the point set that is The farthest point ;

[0076] S3.3. If Of The value is greater than 0.2, it means that Is not a point on the skeleton line and needs to be refined. Calculate And Of , And The mean value of the values, update The coordinate value of, and Of The value is updated to the distance from the Nearest point;

[0077] S3.4. Perform octree filtering on the refined data to remove discrete points and obtain the roughly calculated centerline.

[0078] S4 includes, S4.1. Perform principal component analysis on the neighborhood of each point in the roughly calculated centerline, and calculate the points where the difference between the first principal component and the second principal component is less than the threshold 13 to obtain the cross-region point set;

[0079] S4.2. Use the DBScan algorithm to cluster the cross-region point set, and each intersection is clustered into one class.

[0080] S4 includes, S4.3. According to the structural characteristics of the mine roadway, such as Figure 4 And Figure 5 As shown, define the roadway intersection as Y-shaped and T-shaped. The cross-region point set presents a triangular feature, and obtain three vertices , And ;

[0081] Connect the three obtained vertices into the three sides of a triangle, and obtain the respective center points lmean_01, lmean_12, and lmean_20 of the three sides;

[0082] Extend the vertices to the respective connected centerlines to obtain the respective distal points max_distance0, max_distance1, and max_distance2 of the vertices;

[0083] Connect the distal points to the center points to obtain three angles angle_01, angle_12, and angle_02 that can reflect the intersection type;

[0084] Calculate the difference between the smallest angle and the medium-sized angle. If the difference is less than 10°, it is a T-shaped intersection; otherwise, it is a Y-shaped intersection.

[0085] S4 includes S4.4. For a Y-shaped intersection, the side corresponding to the largest angle is the centerline of the straight lane. Generate a straight line by directly connecting the two distal points corresponding to the largest angle, and evenly generate straight line points according to the point cloud density as the points of the centerline. The side corresponding to the medium angle is the bifurcated lane of the large lane, showing the characteristic of curving outwards. Use the cubic B-spline curve fitting method to calculate the center point of the lane. The side corresponding to the smallest angle is not the centerline, and directly delete the side corresponding to the smallest angle. The refined extraction result is as Figure 6 shown.

[0086] S4 includes S4.5. For a T-shaped intersection, the side corresponding to the largest angle is the centerline of the straight lane. Generate a straight line by directly connecting the two distal points corresponding to the largest angle, and evenly generate straight line points according to the point cloud density as the points of the centerline. Generate a new skeleton line between the third point and this center point. The refined extraction result is as Figure 7 shown. After processing the point sets of different intersection areas respectively, the accurately calculated centerline of the lane is as Figure 8 shown.

[0087] S5 includes S5.1. Mark and sort the obtained center points, store the generated center points in the array , create a new array and initialize it to -1, indicating that the corresponding center point has not been sorted yet, indicating that the corresponding center point enters the sorting sequence, indicating the th element in the array;

[0088] S5.2. Use the minimum spanning tree to connect the center points of the straight lanes and the center points of the intersecting lanes respectively. Each section forms the shortest connected line. Finally, connect the endpoints of the accurately extracted centerline of the intersecting lane with the centerline of the straight lane to generate the complete centerline of the lane.

[0089] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features, and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for calculating the centerline of a complex laneway network based on skeleton shrinkage, characterized in that: include: S1. Point cloud data of underground mine tunnels Establish KDTree for initial contraction to obtain neighborhood distance, read point cloud data point by point and calculate the neighborhood point set of each point, and calculate the surface feature and line feature of each point from the feature value of the neighborhood point set; S2. Iteratively shrink the underground mine tunnel point cloud data, continuously update the coordinate value, plane feature, linear feature and neighborhood range of each point, and obtain the roughly shrunk skeleton point; S3. Refine the skeleton according to the geometric features of each contracted skeleton point, and remove discrete noise points by combining the octree filtering method to obtain the center points of the straight lanes and curved lanes and the point set of the intersection area; S4. By analyzing the morphological characteristics of underground mine tunnels, the Y-shaped and T-shaped tunnel intersections are defined, the principal component method and DBScan algorithm are used to process the intersection area point set, and the intersection type is determined, and the center point of each intersection is refined. S5. smoothly connect the refined calculated center points with other center points, and use the minimum spanning tree to organize the disordered center points, so as to fit the disordered center points into an ordered center line; S4 includes, S4.

1. performing principal component analysis on the neighborhood of each point in the roughly calculated center line, calculating the points where the difference between the first principal component and the second principal component is less than a threshold value of 13 to obtain a set of intersection area points; S4.

2. Use the DBScan algorithm to cluster the intersection area point set, and each intersection is clustered into one category.

2. The method for calculating the centerline of a complex laneway network based on skeleton shrinkage according to claim 1 is characterized in that: S1 includes, S1.

1. For the Point cloud data , , is the total number of point cloud data, based on the distance The distance to the second closest point Set the distance neighborhood D for field search, D is: ； Get the neighborhood point set with D as radius , , , and For the Neighborhood points , and value, is the total number of neighbors.

3. The method for calculating the centerline of a complex laneway network based on skeleton shrinkage according to claim 2 is characterized in that: S1 includes, S1.

2. Calculate the distance mean of the neighborhood range based on the neighborhood point set : ； In the formula, , and They are exist , and The range of directions, and They are exist , and Maximum and minimum values ​​in direction; by Re-search the neighborhood point set for the neighborhood radius and calculate the eigenvalues ​​based on the covariance matrix of the neighborhood point set , and , get the line feature Features of dough : 。 4. The method for calculating the centerline of a complex laneway network based on skeleton shrinkage according to claim 3 is characterized in that: S2 includes, S2.

1. After the initial contraction of the neighborhood-based skeleton, the point cloud data is updated to points with 6 attribute values ; S2.

2. Perform iterative contraction. If Close to 0, If it is close to 1, it means that the current point presents a surface feature rather than a skeleton line. The neighborhood point set requeried with the current point The coordinate mean of the point is updated , and value, , , and Respectively The neighborhood points of the requeried , and value, is the total number of neighborhood points to be requeried: ； In the formula, , and The updated , and value; Update the neighborhood distance: ； In the formula, is the mean distance of the updated neighborhood range; Keep updating , the number of shrinking rounds is 4.

5. The method for calculating the centerline of a complex laneway network based on skeleton shrinkage according to claim 4, characterized in that: S3 includes, S3.

1. analyzing the skeleton points after coarse shrinkage, and establishing KDTree for further refinement of the coarse skeleton line; S3.

2. Re-update of , and Value, according to the point cloud data value, find twice The neighborhood point set of the range, find the distance between the points The farthest point ; S3.

3. If of If the value is greater than 0.2, it means Points that are not on the skeleton line need to be refined and calculated and of , and The mean of the values, update The coordinate value of of The value is updated to The distance to the closest point; S3.

4. Perform octree filtering on the refined data to remove discrete points and obtain a roughly calculated center line.

6. The method for calculating the centerline of a complex laneway network based on skeleton shrinkage according to claim 5, characterized in that: S4 includes, S4.

3. According to the structural characteristics of the mine tunnel, the tunnel intersection is defined as Y-type and T-type, the intersection area point set presents a triangular feature, and the three vertices are obtained. , and ; Connect the three vertices obtained to form the three sides of a triangle, and obtain the center points of the three sides, lmean_01, lmean_12, and lmean_20; Extend the vertices to their respective connected center lines to obtain their respective far end points max_distance0, max_distance1, max_distance2; Connect the far end point and the center point to get three angles angle_01, angle_12, angle_02 that can reflect the type of intersection; Calculate the difference between the smallest angle and the middle angle. If the difference is less than 10°, it is a T-type intersection, otherwise it is a Y-type intersection.

7. The method for calculating the centerline of a complex laneway network based on skeleton shrinkage according to claim 6, characterized in that: S4 includes, S4.

4. For the Y-type intersection, the side corresponding to the maximum angle is the center line of the straight lane. The two far end points corresponding to the maximum angle are directly connected to generate a straight line, and the straight line points are evenly generated as the points of the center line according to the point cloud density. The side corresponding to the middle angle is the branch lane of the large lane, showing the characteristics of arc-shaped projection. The cubic B-spline curve fitting method is used to calculate the center point of the lane. The side corresponding to the minimum angle is not the center line, and the side corresponding to the minimum angle is directly deleted.

8. The method for calculating the centerline of a complex laneway network based on skeleton shrinkage according to claim 7, characterized in that: S4 includes, S4.

5. For a T-intersection, the side corresponding to the maximum angle is the center line of the straight lane. The two far end points corresponding to the maximum angle are directly connected to generate a straight line, and straight line points are uniformly generated as center line points according to the point cloud density, and a new skeleton line is generated between the third point and the center point. After processing different intersection area point sets respectively, the accurately calculated lane center line is obtained.

9. The method for calculating the centerline of a complex laneway network based on skeleton shrinkage according to claim 8, characterized in that: S5 includes, S5.

1. marking and sorting the acquired center points, and storing the generated center points in an array In the example above, create a new array And initialized to -1, express The corresponding center points have not been sorted yet. express The corresponding center point enters the sorting sequence, Indicates the first elements; S5.

2. Use the minimum spanning tree to connect the center points of the straight lanes and the center points of the cross lanes respectively. Each segment constitutes the shortest connected line. Finally, the center line of the cross lane extracted precisely is connected to the center line of the straight lane by endpoints to generate the complete center line of the lane.

Images