Ore vein boundary extraction method, computer device and storage medium

Abstract

The present invention discloses a method for extracting vein boundaries, a computer device and a storage medium, and relates to the technical field of data processing. The method comprises: collecting point cloud data of a mine tunnel by a three-dimensional laser scanner to obtain initial point cloud data; segmenting the initial point cloud data to obtain non-ground point cloud data; gridding the non-ground point cloud data to obtain an initial point cloud grid map; extracting an elevation point cloud image and a reflection intensity point cloud image according to the initial point cloud grid map; based on a maximum entropy threshold, performing point cloud elimination on the initial point cloud grid map according to the elevation point cloud image and the reflection intensity point cloud image to obtain a target point cloud grid map; based on the Otsu threshold method, performing threshold segmentation on the target point cloud grid map according to the reflection intensity point cloud image to obtain a target vein boundary. In this way, the use of reflection intensity information can assist in identifying boundary differences between objects, which helps to improve the accuracy and stability of vein boundary extraction.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to a method for extracting vein boundaries, a computer device, and a storage medium. Background Art

[0002] When the prior art solution extracts vein boundaries, it first collects the projection of the road pothole point cloud in the road plane to obtain a point cloud data set of the potholes; then sequentially extracts sparse boundary points and dense boundary points. This will lead to inaccurate extraction of vein boundaries due to the limitations of mineral identification. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to overcome the deficiencies in the prior art, and provide a method for extracting vein boundaries, a computer device, and a storage medium, which are used to utilize the different laser reflectivities of veins and waste rocks, and adopt a filtering algorithm and threshold segmentation to extract the reflection intensity to identify vein boundaries.

[0004] The present invention provides the following technical solutions:

[0005] In a first aspect, the present invention proposes a method for extracting vein boundaries, including:

[0006] Collect point cloud data of a mine roadway through a three-dimensional laser scanner to obtain initial point cloud data;

[0007] Segment according to the initial point cloud data to obtain non-ground point cloud data;

[0008] Perform grid division on the non-ground point cloud data to obtain an initial point cloud grid map;

[0009] Extract an elevation point cloud image and a reflection intensity point cloud image according to the initial point cloud grid map;

[0010] Based on the maximum entropy threshold, perform point cloud elimination on the initial point cloud grid map according to the elevation point cloud image and the reflection intensity point cloud image to obtain a target point cloud grid map;

[0011] Based on the Otsu threshold method, perform threshold segmentation on the target point cloud grid map according to the reflection intensity point cloud image to obtain a target vein boundary.

[0012] In an embodiment, the segmenting according to the initial point cloud data to obtain non-ground point cloud data includes:

[0013] Denoise the initial point cloud data to obtain target point cloud data;

[0014] Based on the progressive morphological filtering method, segment the target point cloud data according to the height difference threshold to obtain the non-ground point cloud data;

[0015] The expression of the height difference threshold is as follows:

[0016] Wherein, is the height difference threshold under the current window size, is the initial elevation threshold, is the slope parameter, is the maximum height difference threshold, is the th filtering window size, is the size of the divided grid.

[0017] In one embodiment, the denoising of the initial point cloud data to obtain the target point cloud data includes:

[0018] Calibrating the initial point cloud data to obtain the initial point cloud data that is horizontal in the x and y directions;

[0019] Denoising the initial point cloud data that is horizontal in the x and y directions to obtain the target point cloud data.

[0020] In one embodiment, the calibrating of the initial point cloud data to obtain the initial point cloud data that is horizontal in the x and y directions includes:

[0021] Obtaining the measurement values in the stationary state within a preset time period through an inertial measurement unit;

[0022] Determining the gravity direction according to the measurement values;

[0023] Constructing an initial pose based on the Schmidt orthogonalization according to the gravity direction;

[0024] Registering the initial point cloud data according to the initial pose to obtain the initial point cloud data that is horizontal in the x and y directions.

[0025] In one embodiment, the grid division of the non-ground point cloud data to obtain the initial point cloud grid map includes:

[0026] Traversing the position information of each point cloud in the non-ground point cloud data to obtain the coordinate thresholds in each axial direction; and calculating the grid dimension information according to the coordinate thresholds;

[0027] Based on a preset step size, dividing the non-ground point cloud data according to the grid dimension information to obtain the initial point cloud grid map.

[0028] In one embodiment, the extracting of the elevation point cloud image and the reflection intensity point cloud image according to the initial point cloud grid map includes:

[0029] Calculate the average value of the point cloud height values in each grid of the initial point cloud grid map, and construct the elevation point cloud image according to the average value of the point cloud height values;

[0030] Calculate the average value of the point cloud intensity values in each grid of the initial point cloud grid map, and construct the reflection intensity point cloud image according to the average value of the point cloud intensity values.

[0031] In one embodiment, the maximum entropy threshold includes a first maximum entropy threshold and a second maximum entropy threshold. Based on the maximum entropy threshold, the initial point cloud grid map is subjected to point cloud rejection according to the elevation point cloud image and the reflection intensity point cloud image to obtain a target point cloud grid map, including:

[0032] Process the elevation point cloud image and the reflection intensity point cloud image respectively according to the second-order differential operator to obtain an elevation gradient image and a reflection intensity gradient image;

[0033] Calculate the entropy of the elevation gradient image to obtain the first maximum entropy threshold; calculate the entropy of the reflection intensity gradient image to obtain the second maximum entropy threshold;

[0034] Perform point cloud rejection on the initial point cloud grid map according to the first maximum entropy threshold and the second maximum entropy threshold to obtain the target point cloud grid map.

[0035] In one embodiment, based on the Otsu threshold method, the target point cloud grid map is subjected to threshold segmentation according to the reflection intensity point cloud image to obtain a target vein boundary, including:

[0036] Calculate the global threshold of the reflection intensity gradient image based on the Otsu threshold method, and perform threshold segmentation on the target point cloud grid map according to the global threshold to obtain an initial vein boundary;

[0037] Perform filling processing on the initial vein boundary based on the region growing algorithm to obtain a candidate vein boundary;

[0038] Process the candidate vein boundary based on the Euclidean clustering algorithm to obtain the target vein boundary.

[0039] In a second aspect, the present invention proposes a computer device, including a memory and a processor, where the memory stores a computer program, and when the computer program is executed by the processor, the vein boundary extraction method as described in the first aspect is implemented.

[0040] In a third aspect, the present invention proposes a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the vein boundary extraction method as described in the first aspect is implemented.

[0041] The vein boundary extraction method, computer device and storage medium disclosed by the present invention collect point cloud data of a mine roadway through a three-dimensional laser scanner to obtain initial point cloud data; segment the initial point cloud data to obtain non-ground point cloud data; perform grid division on the non-ground point cloud data to obtain an initial point cloud grid map; extract an elevation point cloud image and a reflection intensity point cloud image according to the initial point cloud grid map; based on the maximum entropy threshold, perform point cloud elimination on the initial point cloud grid map according to the elevation point cloud image and the reflection intensity point cloud image to obtain a target point cloud grid map; based on the Otsu threshold method, perform threshold segmentation on the target point cloud grid map according to the reflection intensity point cloud image to obtain a target vein boundary. In this way, while the elevation information provides an elevation basis, the reflection intensity information can assist in identifying the boundary differences between objects, which helps to improve the accuracy and stability of vein boundary extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the present invention, the following will briefly introduce the drawings required in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as limiting the protection scope of the present invention. In each drawing, similar components are numbered similarly.

[0043] Figure 1 FIG. 1 shows a flowchart of the vein boundary extraction method proposed in this embodiment;

[0044] Figure 2 FIG. 2 shows another flowchart of the vein boundary extraction method proposed in this embodiment;

[0045] Figure 3 FIG. 3 shows yet another flowchart of the vein boundary extraction method proposed in this embodiment;

[0046] Figure 4 FIG. 4 shows still another flowchart of the vein boundary extraction method proposed in this embodiment;

[0047] Figure 5 FIG. 5 shows yet another flowchart of the vein boundary extraction method proposed in this embodiment;

[0048] Figure 6 FIG. 6 shows a structural diagram of the vein boundary extraction system proposed in this embodiment.

[0049] DESCRIPTION OF THE REFERENCE NUMERALS

[0050] 600 - vein boundary extraction system; 601 - acquisition module; 602 - segmentation module; 603 - division module; 604 - extraction module; 605 - elimination module; 606 - determination module. Detailed implementation mode

[0051] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments.

[0052] Generally, the components of the embodiments of the present invention described and illustrated herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0053] In the following, the terms "including", "having" and their cognates that can be used in various embodiments of the present invention are only intended to represent specific features, numbers, steps, operations, elements, components or combinations of the foregoing items, and should not be construed as first excluding the existence of one or more other features, numbers, steps, operations, elements, components or combinations of the foregoing items or increasing the possibility of one or more features, numbers, steps, operations, elements, components or combinations of the foregoing items.

[0054] In addition, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0055] Unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the various embodiments of the present invention belong. The terms (such as those defined in a commonly used dictionary) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having an idealized meaning or an overly formal meaning unless clearly defined in the various embodiments of the present invention.

[0056] Embodiment 1

[0057] The embodiment of the present disclosure provides a method for extracting vein boundaries, which is used to collect mine point cloud data, and uses the different laser reflectivities of veins and waste rocks to extract the reflection intensity by using a filtering algorithm and threshold segmentation to identify the vein boundaries.

[0058] Please refer to Figure 1 , a method for extracting vein boundaries includes steps S101 to S106, and the following will explain each step in detail.

[0059] Step S101, collect the point cloud data of the mine roadway through a three-dimensional laser scanner to obtain the initial point cloud data.

[0060] In this embodiment, the roadway point cloud data is collected by a handheld three-dimensional laser scanner, and a large amount of roadway point cloud data can be collected in a short time to obtain the initial point cloud data.

[0061] Step S102: Segment according to the initial point cloud data to obtain non-ground point cloud data.

[0062] In this embodiment, by segmenting the initial point cloud data, non-ground point cloud data can be obtained, and this non-ground point cloud data contains ore veins, so that the ore vein boundaries can be extracted from the non-ground point cloud data subsequently.

[0063] Please refer to Figure 2 , in a specific embodiment, step S102 includes steps S1021 to S1022, and the following is a detailed description of each step.

[0064] Step S1021: Denoise the initial point cloud data to obtain target point cloud data.

[0065] In this embodiment, due to the complex underground environment with problems such as water mist and dust, noise points will be generated during the process of the lidar collecting point cloud data, which will inevitably affect the extraction of ore vein boundaries. Therefore, it is necessary to use a deep learning network to remove the noise points and outliers in the initial point cloud data to obtain the target point cloud data.

[0066] Please refer to Figure 3 , in a specific embodiment, step S1021 includes steps S301 to S302, and the following is a detailed description of each step.

[0067] Step S301: Calibrate the initial point cloud data to obtain the initial point cloud data that is horizontal in the x and y directions.

[0068] In this embodiment, since the initial point cloud data of the mine roadway is not horizontal in the x and y directions, this will inevitably lead to a deterioration in the segmentation quality of the ground point cloud and the non-ground point cloud. Therefore, it is necessary to calibrate the initial point cloud data to obtain the initial point cloud data that is horizontal in the x and y directions.

[0069] In a specific embodiment, step S301 includes: obtaining the measurement values in the stationary state within a preset time period through an inertial measurement unit; determining the gravity direction according to the measurement values; constructing an initial pose based on Schmidt orthogonalization according to the gravity direction; and registering the initial point cloud data according to the initial pose to obtain the initial point cloud data that is horizontal in the x and y directions.

[0070] In this embodiment, the measured values within a preset time period when the inertial measurement unit (IMU) is stationary are obtained, which usually include the data of a three-axis accelerometer and a three-axis gyroscope. Since the accelerometer mainly measures the gravitational acceleration when the IMU is in a stationary state (assuming no other external forces), these measured values can be used to determine the direction of gravity.

[0071] Align the measured values with the local direction of gravity. When the direction of gravity is determined and aligned with the world coordinate system, the Schmidt orthogonalization process can be used to construct an orthogonal basis, which can be used to define the initial attitude of the IMU. Schmidt orthogonalization is a method for orthogonalizing a set of linearly independent vectors and can be used to construct two horizontal direction vectors (usually the x-axis and y-axis) that are orthogonal to the direction of gravity.

[0072] Exemplarily, taking the preset time period as 2 s as an example, let be the acceleration information measured by the IMU. At this time, the measured value on is the gravitational acceleration, and its magnitude is equal to the gravitational acceleration , and the direction is opposite to . Calculate the mean value of the accelerometer within 2 s to obtain , and take out the axial direction value to construct a non-zero vector , where represents the normalized result. Arbitrarily take non-zero vectors , , and construct a set of orthogonal bases through Schmidt orthogonalization:

[0073]

[0074]

[0075]

[0076] Through , , and construct the initial pose.

[0077] Furthermore, using the initial pose, convert the first-frame point cloud data into the global coordinate system to construct the first-frame point cloud map. In this way, the first-frame point cloud serves as the starting point of the point cloud map.

[0078] Combine the second-frame point cloud data in the initial point cloud data with the IMU data at the corresponding moment to construct the second-frame point cloud map. The IMU data can provide information about the sensor acceleration and angular velocity, which is used to estimate the relative pose change between consecutive frames.

[0079] Register the first frame of point cloud map with the first frame of point cloud map, that is, apply the pose of the estimated second frame of point cloud to its point cloud data and transform it into the global coordinate system.

[0080] Use point cloud registration algorithms (such as ICP, NDT, etc.) to further align the second frame of point cloud with the first frame of point cloud map, so as to reduce the pose estimation error caused by IMU integration error or other factors. Finally, obtain the initial point cloud data that is horizontal in the x and y directions. The initial point cloud data that is horizontal in the x and y directions is a set of point cloud data that is distributed in the x and y directions and has relatively small changes in the z direction.

[0081] Step S302: Denoise the initial point cloud data that is horizontal in the x and y directions to obtain the target point cloud data.

[0082] In this embodiment, based on the initial point cloud data that is horizontal in the x and y directions, add Gaussian noise with different levels to construct a data set; construct a deep learning network structure, use PointCleanNet as the network structure, load the pre-trained model, and use the initial point cloud data that is horizontal in the x and y directions to retrain the pre-trained model to prevent overfitting of the PointCleanNet network. The expression of the loss function used in the training process is as follows:

[0083]

[0084] In the formula is the noise point, is the estimated denoised point , is the neighborhood of the estimated denoised point, is a parameter, generally set to 0.99.

[0085] Send the collected point cloud data of the mine roadway into the trained PointCleanNet network to obtain the denoised target point cloud data.

[0086] Please refer to Figure 2 again, step S1022, based on the progressive morphological filtering method, segment the target point cloud data according to the height difference threshold to obtain the non-ground point cloud data.

[0087] In this embodiment, adopt the progressive morphological filtering method. By continuously increasing the filtering window and setting the height difference threshold, retain the points smaller than the height difference threshold, and segment the target point cloud data into ground point cloud data and non-ground point cloud data. The height difference threshold can be expressed as:

[0088]

[0089] In the formula, is the height difference threshold for the current window size, is the initial elevation threshold, is the slope parameter, is the maximum height difference threshold, is the th filter window size, is the size of the divided grid, that is, the size of the grid when loading irregular point clouds into a regular grid structure.

[0090] Please refer to Figure 1 again, step S103, perform grid division on the non-ground point cloud data to obtain an initial point cloud grid map.

[0091] In this embodiment, by dividing the non-ground point cloud data into grids, each grid contains only a limited number of point cloud data, thereby reducing the overall data volume.

[0092] In a specific embodiment, step S103 includes: traversing the position information of each point cloud in the non-ground point cloud data to obtain coordinate thresholds in each axis; and calculating grid dimension information according to the coordinate thresholds; based on a preset step size, perform grid division on the non-ground point cloud data according to the grid dimension information to obtain the initial point cloud grid map.

[0093] In this embodiment, traverse the position information of each point cloud in the non-ground point cloud data , to obtain coordinate thresholds in each axis. The coordinate thresholds include the maximum values in the direction, direction and direction, , and the minimum values , , .

[0094] Calculate grid dimension information according to the coordinate thresholds; and based on a preset step size, perform grid division on the non-ground point cloud data according to the grid dimension information to obtain an initial point cloud grid map.

[0095] Exemplarily, set the preset step size to , then the index of any point cloud in the grid can be expressed as , where can be calculated by the following formula:

[0096]

[0097] Please refer to Figure 1, step S104, extract the elevation point cloud image and the reflection intensity point cloud image according to the initial point cloud raster map.

[0098] In this embodiment, the elevation point cloud image and the reflection intensity point cloud image are extracted according to the initial point cloud raster map. Among them, the elevation point cloud image mainly reflects the height information of the object surface, and the reflection intensity point cloud image mainly reflects the reflection intensity information of the object surface to the laser. By utilizing the different reflectivities of the ore vein and the waste rock to the laser, the extraction of the reflection intensity can reduce the inaccurate extraction of the ore vein boundary caused by the limitations of mineral identification.

[0099] In a specific embodiment, step S104 includes: calculating the average value of the point cloud height values in each grid of the initial point cloud raster map, and constructing the elevation point cloud image according to the average value of the point cloud height values; calculating the average value of the point cloud intensity values in each grid of the initial point cloud raster map, and constructing the reflection intensity point cloud image according to the average value of the point cloud intensity values.

[0100] In this embodiment, traverse each grid in the initial point cloud raster map, and calculate the average value of all point cloud height values in each grid , and use the average value of the point cloud height values to replace all the point cloud height values in the grid to form the elevation point cloud image. Correspondingly, calculate the average value of all point cloud intensity values in each grid , and use the average value of the point cloud intensity values to replace all the intensity values of the point clouds in the grid to form the reflection intensity point cloud image.

[0101] Among them, , .

[0102] In the formula, is the number of point clouds in the grid, , is the intensity value corresponding to each point cloud in the grid.

[0103] Please refer to Figure 1 again, step S105, based on the maximum entropy threshold, perform point cloud elimination on the initial point cloud raster map according to the elevation point cloud image and the reflection intensity point cloud image, and obtain the target point cloud raster map.

[0104] In this embodiment, calculate the maximum entropy threshold according to the elevation point cloud image and the reflection intensity point cloud image, and perform elimination on the point cloud data with a smaller slope and a smaller intensity in the initial point cloud raster map according to the maximum entropy threshold, so as to obtain the target point cloud raster map.

[0105] Through the maximum entropy threshold of the elevation point cloud image, ground points and non-ground points can be accurately identified, thereby removing ground points irrelevant to the vein and improving the accuracy of the point cloud data. The reflection intensity point cloud image provides the reflection intensity information of the point cloud. Utilizing the maximum entropy threshold of the reflection intensity point cloud image helps to further distinguish point clouds of different materials and attributes, thereby removing interference points such as vegetation and soil.

[0106] Please refer to Figure 4 , in a specific embodiment, the maximum entropy threshold includes a first maximum entropy threshold and a second maximum entropy threshold, and step S105 includes steps S1051 to S1053. The following will explain each step in detail.

[0107] Step S1051, process the elevation point cloud image and the reflection intensity point cloud image respectively according to the second-order differential operator to obtain an elevation gradient image and a reflection intensity gradient image.

[0108] In this embodiment, traverse the target point cloud raster map, and use the second-order differential (Laplacian) operator to process the elevation point cloud image and the reflection intensity point cloud image respectively to obtain an elevation gradient image and a reflection intensity gradient image.

[0109] Step S1052, calculate the entropy of the elevation gradient image to obtain the first maximum entropy threshold; calculate the entropy of the reflection intensity gradient image to obtain the second maximum entropy threshold.

[0110] In this embodiment, calculate the entropy of the elevation gradient image to obtain the maximum entropy threshold of the elevation gradient image, that is, the first maximum entropy threshold; calculate the entropy of the reflection intensity gradient image to obtain the maximum entropy threshold of the reflection intensity gradient image, that is, the second maximum entropy threshold.

[0111] Step S1053, perform point cloud removal on the initial point cloud raster map according to the first maximum entropy threshold and the second maximum entropy threshold to obtain the target point cloud raster map.

[0112] In this embodiment, perform a binarization operation on the initial point cloud raster map according to the first maximum entropy threshold, assign pixels greater than the first maximum entropy threshold to 1, and assign pixels less than the first maximum entropy threshold to 0, and remove the point cloud data with a small slope (i.e., pixel value of 0) in the initial point cloud raster map.

[0113] Furthermore, perform a binarization operation on the initial point cloud raster map after removal according to the second maximum entropy threshold, assign pixels greater than the second maximum entropy threshold to 1, and assign pixels less than the second maximum entropy threshold to 0, and remove the point cloud data with a small intensity (i.e., pixel value of 0) in the initial point cloud raster map, and finally obtain the target point cloud raster map.

[0114] Please refer to again Figure 1 In step S106, based on the Otsu threshold method, the target point cloud raster map is threshold segmented according to the reflection intensity point cloud image to obtain the target vein boundary.

[0115] In this embodiment, by calculating the global threshold based on the Otsu threshold method according to the reflection intensity point cloud image and segmenting the target point cloud raster map according to the global threshold, a more accurate target vein boundary can be obtained.

[0116] It should be noted that the processed point cloud data and the target vein boundary can be visually displayed in three dimensions, facilitating the user to intuitively analyze the vein structure and boundary information.

[0117] Please refer to Figure 5 In a specific embodiment, step S106 includes steps S1061 to S1063, and the following is a detailed description of each step.

[0118] In step S1061, based on the Otsu threshold method, the global threshold of the reflection intensity gradient image is calculated, and the target point cloud raster map is threshold segmented according to the global threshold to obtain the initial vein boundary.

[0119] In this embodiment, the global threshold of the reflection intensity gradient image is calculated using the Otsu threshold method, and the target point cloud raster map is threshold segmented according to the global threshold, that is, the pixel points greater than the global threshold are identified as the initial vein boundary, and the initial vein boundary is a rough vein boundary. Through the optimized threshold division method, the detailed information of the vein boundary is retained to the greatest extent, while effectively reducing the interference of background noise.

[0120] In step S1062, the initial vein boundary is filled based on the region growing algorithm to obtain the candidate vein boundary.

[0121] In this embodiment, the region growing algorithm is used to fill the initial vein boundary, thereby generating a continuous candidate vein boundary. By using the spatial connectivity of the boundary pixel points, the phenomenon of broken and discontinuous edges is effectively eliminated, providing more complete boundary information for subsequent processing.

[0122] In step S1063, the candidate vein boundary is processed based on the Euclidean clustering algorithm to obtain the target vein boundary.

[0123] In this embodiment, by applying the Euclidean clustering algorithm to the point cloud data within the candidate vein boundary for processing, the point cloud data can be grouped according to its geometric structure, and the spatial distribution characteristics of the vein region can be identified, thereby obtaining the target vein boundary. This clustering-based processing method can improve the accuracy and reliability of vein data extraction, laying a solid foundation for vein morphology analysis and subsequent three-dimensional modeling.

[0124] The vein boundary extraction method proposed in this embodiment collects the point cloud data of the mine roadway through a 3D laser scanner to obtain the initial point cloud data; performs segmentation on the basis of the initial point cloud data to obtain non-ground point cloud data; conducts grid division on the non-ground point cloud data to obtain the initial point cloud grid map; extracts the elevation point cloud image and the reflection intensity point cloud image according to the initial point cloud grid map; based on the maximum entropy threshold, performs point cloud elimination on the initial point cloud grid map according to the elevation point cloud image and the reflection intensity point cloud image to obtain the target point cloud grid map; based on the Otsu threshold method, performs threshold segmentation on the target point cloud grid map according to the reflection intensity point cloud image to obtain the target vein boundary. In this way, while the elevation information provides an elevation basis, the reflection intensity information can assist in identifying the boundary differences between objects, which helps to improve the accuracy and stability of vein boundary extraction.

[0125] Embodiment 2

[0126] In addition, an embodiment of the present disclosure provides a vein boundary extraction system 600. Please refer to Figure 6 , and this system includes:

[0127] A collection module 601, configured to collect the point cloud data of the mine roadway through a 3D laser scanner to obtain the initial point cloud data;

[0128] A segmentation module 602, configured to perform segmentation on the basis of the initial point cloud data to obtain non-ground point cloud data;

[0129] A division module 603, configured to conduct grid division on the non-ground point cloud data to obtain the initial point cloud grid map;

[0130] An extraction module 604, configured to extract the elevation point cloud image and the reflection intensity point cloud image according to the initial point cloud grid map;

[0131] An elimination module 605, configured to perform point cloud elimination on the initial point cloud grid map based on the maximum entropy threshold according to the elevation point cloud image and the reflection intensity point cloud image to obtain the target point cloud grid map;

[0132] A determination module 606, configured to perform threshold segmentation on the target point cloud grid map based on the Otsu threshold method according to the reflection intensity point cloud image to obtain the target vein boundary.

[0133] Optionally, the segmentation module 602 is configured to denoise the initial point cloud data to obtain target point cloud data; based on the progressive morphological filtering method, perform segmentation on the target point cloud data according to the elevation difference threshold to obtain the non-ground point cloud data; the expression of the elevation difference threshold is: ; where is the height difference threshold for the current window size, is the initial elevation threshold, is the slope parameter, is the maximum height difference threshold, is the th filtering window size, is the size of the divided grid.

[0134] Optionally, the segmentation module 602 is used to calibrate the initial point cloud data to obtain the initial point cloud data that is horizontal in the x and y directions; denoise the initial point cloud data that is horizontal in the x and y directions to obtain the target point cloud data.

[0135] Optionally, the segmentation module 602 is used to obtain the measurement values in the stationary state within a preset time period through an inertial measurement unit; determine the gravity direction according to the measurement values; construct an initial pose based on the Schmidt orthogonalization according to the gravity direction; register the initial point cloud data according to the initial pose to obtain the initial point cloud data that is horizontal in the x and y directions.

[0136] Optionally, the division module 603 is used to traverse the position information of each point cloud in the non-ground point cloud data to obtain the coordinate thresholds in each axis; calculate the grid dimension information according to the coordinate thresholds; perform grid division on the non-ground point cloud data based on a preset step size according to the grid dimension information to obtain the initial point cloud grid map.

[0137] Optionally, the extraction module 604 is used to calculate the average value of the point cloud height values in each grid of the initial point cloud grid map, construct the elevation point cloud image according to the average value of the point cloud height values; calculate the average value of the point cloud intensity values in each grid of the initial point cloud grid map, and construct the reflection intensity point cloud image according to the average value of the point cloud intensity values.

[0138] Optionally, the maximum entropy threshold includes a first maximum entropy threshold and a second maximum entropy threshold. The rejection module 605 is used to process the elevation point cloud image and the reflection intensity point cloud image respectively according to the second-order differential operator to obtain an elevation gradient image and a reflection intensity gradient image; calculate the entropy of the elevation gradient image to obtain the first maximum entropy threshold; calculate the entropy of the reflection intensity gradient image to obtain the second maximum entropy threshold; perform point cloud rejection on the initial point cloud grid map according to the first maximum entropy threshold and the second maximum entropy threshold to obtain the target point cloud grid map.

[0139] Optionally, a determination module 606 is configured to calculate a global threshold of the reflection intensity gradient image based on the Otsu threshold method, perform threshold segmentation on the target point cloud raster map according to the global threshold to obtain an initial vein boundary; perform filling processing on the initial vein boundary based on a region growing algorithm to obtain a candidate vein boundary; and process the candidate vein boundary based on an Euclidean clustering algorithm to obtain the target vein boundary.

[0140] The system provided by the embodiments of the present disclosure can execute the steps of the vein boundary extraction method provided in Embodiment 1. To avoid repetition, details are not described herein again.

[0141] The vein boundary extraction system proposed in this embodiment collects point cloud data of a mine roadway through a 3D laser scanner to obtain initial point cloud data; performs segmentation on the initial point cloud data to obtain non-ground point cloud data; performs grid division on the non-ground point cloud data to obtain an initial point cloud raster map; extracts an elevation point cloud image and a reflection intensity point cloud image according to the initial point cloud raster map; based on the maximum entropy threshold, performs point cloud rejection on the initial point cloud raster map according to the elevation point cloud image and the reflection intensity point cloud image to obtain a target point cloud raster map; based on the Otsu threshold method, performs threshold segmentation on the target point cloud raster map according to the reflection intensity point cloud image to obtain a target vein boundary. In this way, while the elevation information provides an elevation basis, the reflection intensity information can assist in identifying boundary differences between objects, which helps to improve the accuracy and stability of vein boundary extraction.

[0142] Embodiment 3

[0143] In addition, an embodiment of the present disclosure provides a computer device, including a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the vein boundary extraction method described in Embodiment 1 is implemented.

[0144] The device provided by the embodiments of the present disclosure can execute the steps of the vein boundary extraction method provided in Embodiment 1. To avoid repetition, details are not described herein again.

[0145] Embodiment 4

[0146] An embodiment of the present disclosure proposes a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the vein boundary extraction method described in Embodiment 1 of this embodiment is implemented.

[0147] In this embodiment, the computer-readable storage medium can be a read-only memory (ROM for short), a random access memory (RAM for short), a magnetic disk, or an optical disc, etc.

[0148] The computer-readable storage medium provided in this embodiment can implement the vein boundary extraction method provided in Embodiment 1. To avoid repetition, it will not be elaborated here.

[0149] In all the examples shown and described here, any specific value should be construed as merely exemplary and not as a limitation. Therefore, other examples of the exemplary embodiments may have different values.

[0150] It should be noted that like reference numerals and letters denote like items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0151] The above-described embodiments merely represent several implementation manners of the present invention. The description thereof is relatively specific and detailed, but it should not be construed as a limitation to the scope of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention.

Claims

1. A method for extracting a mineral vein boundary, characterized in that: include: The point cloud data of the mine tunnel is collected by a 3D laser scanner to obtain initial point cloud data; Segmenting the initial point cloud data to obtain non-ground point cloud data; Performing grid division on the non-ground point cloud data to obtain an initial point cloud grid map; Extracting an elevation point cloud image and a reflection intensity point cloud image according to the initial point cloud grid map; Based on the maximum entropy threshold, point cloud removal is performed on the initial point cloud grid map according to the elevation point cloud image and the reflection intensity point cloud image to obtain a target point cloud grid map; Based on the Otsu threshold method, threshold segmentation is performed on the target point cloud grid map according to the reflection intensity point cloud image to obtain the target vein boundary; The maximum entropy threshold includes a first maximum entropy threshold and a second maximum entropy threshold, the first maximum entropy threshold is the maximum entropy threshold of the elevation gradient image, and the second maximum entropy threshold is the maximum entropy threshold of the reflection intensity gradient image; The method of performing point cloud elimination on the initial point cloud grid map based on the maximum entropy threshold and according to the elevation point cloud image and the reflection intensity point cloud image to obtain a target point cloud grid map includes: The elevation point cloud image and the reflection intensity point cloud image are processed according to a second-order differential operator to obtain an elevation gradient image and a reflection intensity gradient image; Calculating the entropy of the elevation gradient image to obtain the first maximum entropy threshold; calculating the entropy of the reflection intensity gradient image to obtain the second maximum entropy threshold; Point cloud culling is performed on the initial point cloud grid map according to the first maximum entropy threshold and the second maximum entropy threshold to obtain the target point cloud grid map.

2. The method for extracting mineral vein boundaries according to claim 1, characterized in that: The step of segmenting the initial point cloud data to obtain non-ground point cloud data includes: De-noising the initial point cloud data to obtain target point cloud data; Based on a progressive morphological filtering method, the target point cloud data is segmented according to a height difference threshold to obtain the non-ground point cloud data; The expression of the height difference threshold is: in, is the height difference threshold under the current window size, is the initial elevation threshold, is the slope parameter, For the Secondary filter window size, is the size of the grid; If the height difference threshold under the current window size is calculated according to the expression of the height difference threshold Greater than the maximum height difference threshold , then the height difference threshold under the current window size The final value of is the maximum height difference threshold .

3. The method for extracting the mineral vein boundary according to claim 2, characterized in that: The denoising of the initial point cloud data to obtain target point cloud data includes: Calibrating the initial point cloud data to obtain horizontal initial point cloud data in the x and y directions; The initial point cloud data horizontally in the x and y directions is denoised to obtain the target point cloud data.

4. The method for extracting mineral vein boundaries according to claim 3, characterized in that: The step of calibrating the initial point cloud data to obtain horizontal initial point cloud data in the x and y directions includes: Obtaining measurement values ​​in a stationary state within a preset time period through an inertial measurement unit; determining a direction of gravity based on the measurements; Based on Schmidt orthogonalization, construct an initial pose according to the gravity direction; The initial point cloud data is registered according to the initial position and posture to obtain the initial point cloud data horizontal in the x and y directions.

5. The method for extracting mineral vein boundaries according to claim 1, characterized in that: The step of performing grid division on the non-ground point cloud data to obtain an initial point cloud grid map includes: Traversing the position information of each point cloud in the non-ground point cloud data to obtain a coordinate threshold in each axis; and calculating grid dimension information according to the coordinate threshold; Based on a preset step size, the non-ground point cloud data is grid-divided according to the grid dimension information to obtain the initial point cloud grid map.

6. The method for extracting mineral vein boundaries according to claim 1, characterized in that: The step of extracting an elevation point cloud image and a reflection intensity point cloud image according to the initial point cloud grid map comprises: Calculating the average value of the point cloud height values ​​in each grid in the initial point cloud grid map, and constructing the elevation point cloud image according to the average value of the point cloud height values; The average value of the point cloud intensity values ​​in each grid in the initial point cloud grid map is calculated, and the reflection intensity point cloud image is constructed according to the average value of the point cloud intensity values.

7. The method for extracting mineral vein boundaries according to claim 1, characterized in that: The method of performing threshold segmentation on the target point cloud grid map based on the reflection intensity point cloud image to obtain the target mineral vein boundary includes: Calculating a global threshold of the reflection intensity gradient image based on the Otsu threshold method, and performing threshold segmentation on the target point cloud grid map according to the global threshold to obtain an initial vein boundary; Filling the initial vein boundary based on a region growing algorithm to obtain a candidate vein boundary; The candidate vein boundaries are processed based on the Euclidean clustering algorithm to obtain the target vein boundaries.

8. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the method for extracting the mineral vein boundary as claimed in any one of claims 1 to 7 is implemented.

9. A computer-readable storage medium, characterized in that: It stores a computer program, which, when executed by a processor, implements the method for extracting the mineral vein boundary as described in any one of claims 1 to 7.

Images