A method for extracting bench lines in open-pit mines

Abstract

This invention discloses a method for extracting open-pit mine bench lines, relating to the field of slope data acquisition technology, including the following steps: S1, constructing a point cloud model: After obtaining the spatial coordinates of each sampling point on the surface of the open-pit mine platform, multiple points are obtained to form a point cloud model. The beneficial effects of this invention are: This invention extracts open-pit mine bench lines by following the steps of constructing a point cloud model, voxel downsampling, calculating the angle between the normal vector of point O and the unit vector, and calculating the open-pit mine bench lines. The overall process is simple and easy to operate, and the final open-pit mine bench lines are obtained. On the one hand, it improves the accuracy of the final open-pit mine bench line data, and on the other hand, it facilitates the precise location of abnormal links when abnormal situations occur during extraction, shortens the overall workflow, meets the needs of use, reduces the labor intensity of manual operation, and comprehensively improves the overall intelligent management.

Description

Technical Field

[0001] This invention relates to the field of slope data acquisition technology, specifically a method for extracting bench lines in open-pit mines. Background Technology

[0002] In open-pit mining, to facilitate layer-by-layer extraction from top to bottom, the minerals and rocks within the mining area are divided into horizontal layers of a certain thickness. These working faces are called benches. Open-pit mine bench lines play a crucial role in various operations, including open-pit mine surveying, mining design, planning, block design, blasting design, and mine area delineation.

[0003] Currently, the main method used to extract feature lines such as bench lines in open-pit mines during actual production is to manually draw them based on DSM. This process is cumbersome and difficult to master quickly. Furthermore, the lack of multiple processing steps during calculation leads to discrepancies in the final data. It also makes it impossible to pinpoint the problem in any step immediately, requiring the entire process to be restarted if an issue arises. This increases the overall workload, resulting in low efficiency, poor accuracy, and inconvenience for intelligent management.

[0004] Therefore, in order to address the above problems, this invention needs to design a method for extracting bench lines in open-pit mines to solve the aforementioned issues. Summary of the Invention

[0005] The purpose of this invention is to provide a method for extracting bench lines in open-pit mines, in order to solve the problems mentioned in the background art. At present, the main method for extracting feature lines such as bench lines in open-pit mines during actual production is to manually draw them according to the DSM. This process is cumbersome and difficult to master quickly. Moreover, the lack of multiple processing during calculation leads to discrepancies in the final data. It is also impossible to locate the problem in a timely manner, so that once a problem occurs, the whole process needs to be restarted, which increases the overall workload, resulting in low efficiency, poor accuracy, and inconvenience for intelligent management.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for extracting bench lines in open-pit mines, comprising the following steps:

[0007] S1. Constructing a point cloud model: After obtaining the spatial coordinates of each sampling point on the surface of the open-pit mine, multiple points are obtained to form a point cloud model;

[0008] S2, Voxel downsampling: First, the point cloud model is divided into voxels, and the corresponding sampling mode is selected to realize the downsampling of the point cloud model;

[0009] S3. Calculate the angle between the normal vector of point O and the unit vector: Calculate the angle between the normal vector of point O and the unit vector, compare it with the threshold, until the required slope feature point O is obtained, and proceed to the next step.

[0010] S4. Calculate the open-pit mine bench line: The point O, which is a slope feature point in step S3, is segmented by the region growing algorithm, and the contour of the segmented feature points is extracted by the concave hull algorithm. The extracted contour feature points are simulated by Bézier curves to obtain the final open-pit mine bench line.

[0011] Preferably, the construction of the point cloud model in step S1 includes the following specific steps:

[0012] S101. The obtained points are processed in sequence using bilateral filtering, Gaussian filtering, conditional filtering, direct filtering, and random sample consistency filtering.

[0013] S102. When the relative positions of point clouds are completely unknown, the point clouds are registered using a coarse-fine registration method to provide good initial values ​​for fine registration. Based on the coarse registration, fine registration is used to minimize the spatial position differences between point clouds and then higher-level processing is performed sequentially.

[0014] S103. After completing the low-level processing of step S101 and the high-level processing of step S102, the point cloud is compressed, summarized, optimized and rendered to construct a point cloud model.

[0015] Preferably, the voxel division in step S2 includes voxel center downsampling and voxel center downsampling;

[0016] The centroid downsampling of the voxel is achieved by calculating the centroid of a non-empty voxel and replacing all points within that voxel, thus realizing the downsampling of the point cloud model.

[0017] The voxel center downsampling method calculates the center of a non-empty voxel to replace all points within that voxel, thereby achieving downsampling of the point cloud model.

[0018] Preferably, the calculation of the angle between the normal vector of point O and the unit vector in step S3 includes the following specific steps:

[0019] S301. Take the downsampled data from step S2, set them as points O in sequence, and record the coordinate values ​​of the selected points O.

[0020] S302. Calculate the angle between the normal vector of point O and the unit vector, and compare it with the threshold.

[0021] S303. If the condition is not met, it means that point O is not a slope feature point. Repeat the above operation to reselect point O. If the condition is met, it means that point O is a slope feature point. Proceed to the next step.

[0022] Preferably, the region growing algorithm segmentation process in step S4 includes the following specific steps:

[0023] Find all points O in the array and erode each point O into a single pixel;

[0024] The pixels are plotted in coordinates to form an image. If the input image satisfies the given properties at the coordinates of that point, the process continues to the next step; otherwise, the point O is reselected. Finally, each connected point O in different regions is marked to form the final segmented image.

[0025] Preferably, the concave hull algorithm extracting the contour of segmentation feature points in step S4 includes the following specific steps:

[0026] Calculate the center point R of discrete points in the segmented image, find the initial boundary point A that is closest to point R, and store it in the boundary point set. Find a point X such that the angle between vector RA and vector RX when rotated counterclockwise is minimized. Calculate the angle between vector RA and any vector RX using the angle formula. Stop iterating when the selected point X coincides with the initial boundary point A, and proceed to the next step.

[0027] Preferably, step S4, which involves simulating contour feature points using Bézier curves, includes the following specific steps:

[0028] Select the corresponding number of data points according to the needs of the simulation. The curve passes through the start and end points inside the data points and is tangent to the start and end points of the feature polygon. The midpoint pulls the curve toward itself.

[0029] Select the appropriate number of control points according to the needs of the simulation. When the number of control points is 1, the curve is a first-order Bézier curve. When the number of control points is 2, the curve is a second-order Bézier curve. When the number of control points is 3, the curve is a third-order Bézier curve.

[0030] Preferably, the extraction method further includes point cloud volume calculation, which includes the following specific steps:

[0031] (1) Construct a two-phase point cloud model based on step S1;

[0032] (2) Calculate the distance between the two point clouds and extract the non-overlapping point cloud;

[0033] (3) Compare with the threshold. If the value is less than the threshold, the point cloud is considered an overlapping point cloud and is removed. If the value is greater than the threshold, the point cloud is considered a non-overlapping point cloud and proceeds to the next step.

[0034] (4) Based on step S4, the non-overlapping point cloud is segmented by the region growing algorithm, the class with a large difference in the number of point clouds between the two periods is removed, each class of point cloud after processing is traversed, the slicing plane and slicing step size are determined, the point cloud is sliced, and the point cloud within the slicing range is projected onto the slicing plane.

[0035] (5) Based on step S4, the projection point cloud contour is extracted by the concave hull algorithm, the point cloud contour area is calculated by Green's formula, the contour area is multiplied by the step size to get the slice area, and the sum of the volumes of each slice is the point cloud volume of that part.

[0036] Preferably, in step (5), before using Green's formula, the outline of the projected point cloud needs to be checked in sequence for curve form, direction and connectivity. When the curve is a closed curve, the direction is a positive direction and the connectivity is partial derivative continuity in the closed region, Green's formula is used for subsequent calculation. If any one of the conditions is not met, the point is determined to be an overlapping point cloud and is removed.

[0037] Compared with the prior art, the beneficial effects of the present invention are:

[0038] This invention extracts open-pit mine bench lines by following the steps of constructing a point cloud model, voxel downsampling, calculating the angle between the normal vector of point O and the unit vector, and calculating the open-pit mine bench lines. The overall process is simple and easy to operate. The final open-pit mine bench lines are obtained by sequentially processing the region growing algorithm, the concave hull algorithm, and the Bézier curve. This improves the accuracy of the final open-pit mine bench line data and facilitates the precise location of abnormal links when anomalies occur during extraction. This makes it easier for staff to view, handle, and summarize the issues. At the same time, it facilitates cloud volume calculation, shortens the overall workflow, meets the needs of users, reduces the labor intensity of manual operation, and comprehensively improves the overall intelligent management. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the overall steps of an open-pit mine bench line extraction method according to an embodiment of the present invention;

[0040] Figure 2 This is a schematic diagram of the calculation steps of an open-pit mine bench line extraction method according to an embodiment of the present invention;

[0041] Figure 3 This is a schematic diagram of the point cloud volume calculation steps of an open-pit mine bench line extraction method according to an embodiment of the present invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Please see Figure 1 , Figure 2 andFigure 3 The present invention provides a technical solution:

[0044] A method for extracting bench lines in open-pit mines includes the following steps:

[0045] S1. Constructing a point cloud model: After obtaining the spatial coordinates of each sampling point on the surface of the open-pit mine, multiple points are obtained to form a point cloud model. The construction of the point cloud model includes the following specific steps: S101. The obtained points are processed sequentially using bilateral filtering, Gaussian filtering, conditional filtering, pass-through filtering, and random sampling consistency filtering. S102. When the relative positions of the point clouds are completely unknown, the point clouds are registered using coarse-fine registration to provide good initial values ​​for fine registration. Based on the coarse registration, fine registration is used to minimize the spatial position differences between the point clouds, and then high-level processing is performed sequentially. S103. The point clouds after completing the low-level processing of step S101 and the high-level processing of step S102 are compressed, summarized, optimized, and rendered to construct a point cloud model.

[0046] S2. Voxel Downsampling: First, the point cloud model is divided into voxels. Then, the corresponding sampling mode is selected to downsample the point cloud model. Voxel division includes voxel centroid downsampling and voxel center downsampling. Voxel centroid downsampling calculates the centroid of a non-empty voxel and replaces all points within that voxel to downsample the point cloud model. Voxel center downsampling calculates the center of a non-empty voxel and replaces all points within that voxel to downsample the point cloud model.

[0047] S3. Calculate the angle between the normal vector of point O and the unit vector: Calculate the angle between the normal vector of point O and the unit vector, compare it with the threshold, until the required slope feature point O is obtained, and proceed to the next step; The specific steps for calculating the angle between the normal vector of point O and the unit vector are as follows: S301. Take the downsampling data from step S2, set it as point O, and record the coordinate values ​​of the selected point O; S302. Calculate the angle between the normal vector of point O and the unit vector, and compare it with the threshold; S303. If the condition is not met, it means that point O is not a slope feature point at this time. Repeat the above operation to reselect point O. If the condition is met, it means that point O is a slope feature point at this time, and proceed to the next step;

[0048] S4. Calculate the open-pit mine bench line: The points O, which are slope feature points in step S3, are segmented using a region growing algorithm. The region growing algorithm segmentation process includes the following specific steps: finding all points O in the array and eroding each point O into a pixel; forming an image from the pixels in coordinates; if the input image satisfies the given attributes at the point coordinates, proceed to the next step; otherwise, reselect point O; finally, mark each connected point O in different regions to form the final segmented image; extract the contours of the segmented feature points using a concave hull algorithm. The concave hull algorithm extracts the contours of the segmented feature points using the following specific steps: calculating the center point R of the discrete points in the segmented image; finding the initial boundary point A closest to point R and storing it in the boundary point set; finding a point X such that vector RA is rotated counterclockwise to the desired direction. The angle between vector RX is minimized. The angle between vector RA and any vector RX is calculated using the angle formula. Iteration continues until the selected point X coincides with the initial boundary point A, at which point the iteration stops and proceeds to the next step. The extracted contour feature points are simulated using Bézier curves. The simulation of contour feature points using Bézier curves includes the following specific steps: Select the corresponding number of data points required for the simulation. The curve passes through the start and end points inside the data points and is tangent to the first and last sides of the feature polygon at the start and end points. The midpoint pulls the curve towards itself. Select the corresponding number of control points required for the simulation. When the number of control points is 1, the curve is a first-order Bézier curve. When the number of control points is 2, the curve is a second-order Bézier curve. When the number of control points is 3, the curve is a third-order Bézier curve, resulting in the final open-pit mine bench line.

[0049] Furthermore, the extraction method also includes point cloud volume calculation, which includes the following specific steps:

[0050] (1) Construct a two-phase point cloud model based on step S1;

[0051] (2) Calculate the distance between the two point clouds and extract the non-overlapping point cloud;

[0052] (3) Compare with the threshold. If the value is less than the threshold, the point cloud is considered an overlapping point cloud and is removed. If the value is greater than the threshold, the point cloud is considered a non-overlapping point cloud and proceeds to the next step.

[0053] (4) Based on step S4, the non-overlapping point cloud is segmented by the region growing algorithm, the class with a large difference in the number of point clouds between the two periods is removed, each class of point cloud after processing is traversed, the slicing plane and slicing step size are determined, the point cloud is sliced, and the point cloud within the slicing range is projected onto the slicing plane.

[0054] (5) Based on step S4, the projection point cloud contour is extracted by the concave hull algorithm, and the area of ​​the point cloud contour is calculated by Green's formula. Before using Green's formula, the curve form, direction and connectivity of the projection point cloud contour need to be checked in turn. When the curve is a closed curve, the direction is a positive direction, and the connectivity is partial derivative continuity in the closed region, the subsequent calculation is performed by Green's formula. If any one of the conditions is not met, the point is determined to be an overlapping point cloud and is removed. The area of ​​the contour is multiplied by the step size to get the area of ​​the slice. The sum of the volumes of each slice is the volume of the point cloud.

[0055] Example

[0056] An open-pit mine bench line extraction method is applied to this field:

[0057] S1. After obtaining the spatial coordinates of each sampling point on the surface of the open-pit mine platform, the open-pit mine workers use bilateral filtering, Gaussian filtering, and random sampling consistency filtering to process the obtained open-pit mine step line points in sequence at low levels. When the relative positions of the point cloud are completely unknown, the point cloud is registered using coarse-fine registration to provide good initial values ​​for fine registration. Based on the coarse registration, fine registration is used to minimize the spatial position differences between the point clouds. Then, high-level processing is carried out in sequence. The open-pit mine step line point cloud after completing low-level and high-level processing is compressed, summarized, optimized, and rendered to construct the open-pit mine step line point cloud model.

[0058] S2. Divide the point cloud model of the open-pit mine step line into voxels, select the corresponding sampling mode, and use the voxel centroid downsampling to replace all points in the voxel by calculating the centroid of the non-empty voxel, and use the voxel center downsampling to replace all points in the voxel by calculating the center of the non-empty voxel, thereby realizing the downsampling of the point cloud model of the open-pit mine step line.

[0059] S3. Set the sampling data of the open-pit mine bench line as point O in sequence, record the coordinate value of the selected point O, calculate the angle between the normal vector of point O and the unit vector, and compare it with the threshold. If the condition is not met, it means that point O is not a slope feature point. Repeat the above operation to reselect point O. If the condition is met, it means that point O is a slope feature point and proceed to the next step.

[0060] S4. Segment the slope feature points O using a region growing algorithm. Find all points O in the array and erode each slope feature point O into a pixel. Construct an image from these pixels in coordinates. If the input image satisfies the given attributes at the point coordinates, proceed to the next step; otherwise, reselect the slope feature points O. Finally, mark each connected slope feature point O in different regions to form the final segmented image. Extract the contours of the segmented feature points using a concave hull algorithm. Calculate the center point R of discrete points in the segmented image, find the initial boundary point A closest to point R, and store it in the boundary point set. Find a point X such that vector RA rotates counterclockwise. The process begins by finding the smallest angle between vectors RX and RA. The angle between RA and any vector RX is then calculated using the angle formula. Iteration continues until the selected point X coincides with the initial boundary point A, at which point the iteration stops. The next step involves simulating the extracted contour feature points using Bézier curves. A corresponding number of data points are selected based on the simulation requirements. The curve passes through the start and end points of the data points. A corresponding number of control points are also selected based on the simulation requirements. When there is one control point, the curve is a first-order Bézier curve; when there are two control points, the curve is a second-order Bézier curve; and when there are three control points, the curve is a third-order Bézier curve. This process yields the final open-pit mine bench line.

[0061] The working principle of this invention is as follows: In use, after obtaining the spatial coordinates of each sampling point on the surface of the open-pit mine platform, the miners use bilateral filtering, Gaussian filtering, conditional filtering, direct-pass filtering, and random sampling consistency filtering to sequentially process the obtained open-pit mine platform line points at low levels. When the relative positions of the point cloud are completely unknown, a coarse-fine registration method is used to register the point cloud, providing good initial values ​​for fine registration. Based on the coarse registration, fine registration is used to minimize the spatial differences between the point clouds, followed by sequential high-level processing. The open-pit mine platform line point cloud, after low-level and high-level processing, is then compressed, summarized, optimized, and rendered to construct the open-pit mine platform. Step line point cloud model; The open-pit mine step line point cloud model is divided into voxels, and the corresponding sampling mode is selected. Voxel centroid downsampling replaces all points within a voxel by calculating the centroid of a non-empty voxel, and voxel center downsampling replaces all points within a voxel by calculating the center of a non-empty voxel, thus realizing the downsampling of the open-pit mine step line point cloud model; The sampled data of the open-pit mine step line are sequentially set as points O, and the coordinate values ​​of the selected points O are recorded. The angle between the normal vector of point O and the unit vector is calculated and compared with a threshold. If the condition is not met, it means that point O is not a slope feature point. The above operation is repeated to reselect points O. If the condition is met, it means that point O is a slope feature point, and the next step is performed;The slope feature points O are segmented using a region growing algorithm. All points O are found in the array, and each feature point O is eroded into a single pixel. The pixels are plotted in coordinates to form an image. If the input image at a given point satisfies a given attribute, the process continues to the next step; otherwise, the slope feature points O are reselected. Finally, each connected slope feature point O is marked in different regions to form the final segmented image. The contours of the segmented feature points are extracted using a concave hull algorithm. The center point R of discrete points in the segmented image is calculated, and the initial boundary point A closest to point R is found and stored in the boundary point set. A point X is found such that the angle between vector RA and vector RX is minimized when RA is rotated counterclockwise. The angle between vector RA and any vector RX is calculated using the angle formula. Iteration continues until the selected point X coincides with the initial boundary point A, then the process stops and proceeds to the next step. The extracted contour feature points are simulated using Bézier curves. The number of data points selected for the simulation is appropriate. The curve passes through the start and end points inside the data points and intersects with the first and last sides of the feature polygon. The curve is drawn towards itself at the starting and ending points, with the intermediate point pulling the curve towards itself. The number of control points is selected according to the simulation requirements. When there is one control point, the curve is a first-order Bézier curve; when there are two control points, the curve is a second-order Bézier curve; and when there are three control points, the curve is a third-order Bézier curve, thus obtaining the final open-pit mine step line. This invention extracts the open-pit mine step line by following the steps of constructing a point cloud model, voxel downsampling, calculating the angle between the normal vector of point O and the unit vector, and calculating the open-pit mine step line. The overall process is simple and easy to operate. Furthermore, the final open-pit mine step line is obtained by sequentially processing the region growing algorithm, the concave hull algorithm, and the Bézier curve. This improves the accuracy of the final open-pit mine step line data and facilitates precise location of abnormalities during extraction, allowing staff to easily view, handle, and summarize the process. It also facilitates cloud volume calculation, shortens the overall workflow, meets the needs of users, reduces the labor intensity of manual operation, and comprehensively improves overall intelligent management.

[0062] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their likenesses.

Claims

1. A method of bench line extraction in an open pit mine, characterized by, The method comprises the following steps: S1, constructing a point cloud model: after obtaining the spatial coordinates of each sampling point on the surface of the open-pit mine bench object, a plurality of points are obtained to form a point cloud model; S2, voxel downsampling: first, the point cloud model is divided into voxels, a corresponding sampling mode is selected, and the point cloud model is downsampled; S3, calculating the angle between the point O normal vector and the unit vector: the angle between the point O normal vector and the unit vector is calculated, and compared with a threshold value until the required slope feature point O is obtained, and the next step is performed; The step S3 of calculating the angle between the point O normal vector and the unit vector comprises the following specific steps: S301, taking the downsampled data in step S2, and sequentially setting it as an O point, and recording the coordinate values of the selected O point; S302, calculating the angle between the point O normal vector and the unit vector, and comparing it with a threshold value; S303, when the condition is not met, it indicates that the point O is a non-slope feature point, and the above operation is repeated to select an O point again, when the condition is met, it indicates that the point O is a slope feature point, and the next step is performed; S4, calculating the open-pit mine bench line: the point O which is the slope feature point in step S3 is processed by region growing algorithm segmentation, the profile of the segmented feature points is extracted by concave hull algorithm, and the extracted profile feature points are simulated by Bezier curve to obtain the final open-pit mine bench line.

2. The bench line extraction method of claim 1, wherein: The step S1 of constructing a point cloud model comprises the following specific steps: S101, sequentially processing the obtained points in a low level by using the ways of bilateral filtering, Gaussian filtering, conditional filtering, straight-through filtering and random sample consensus filtering; S102, when the relative positions of the point clouds are completely unknown, the point clouds are registered by coarse-fine matching to provide a good initial value for fine registration, and the spatial position difference between the point clouds is minimized by fine registration for high-level processing; S103, after the low-level processing of step S101 and the high-level processing of step S102, the point clouds are compressed, summarized, optimized and rendered to construct a point cloud model.

3. The bench line extraction method of claim 1, wherein: The voxel division in step S2 comprises voxel center downsampling and voxel center downsampling; The voxel center downsampling replaces all points in the voxel with the center of the non-empty voxel to realize the downsampling of the point cloud model; The voxel center downsampling replaces all points in the voxel with the center of the non-empty voxel to realize the downsampling of the point cloud model.

4. The bench line extraction method of claim 1, wherein: The region growing algorithm segmentation in step S4 comprises the following specific steps: All points O in the array are found, and each point O is eroded into a pixel; In the coordinates, the pixels form an image, if the input image satisfies the given properties at the point coordinates, the next step is continued, otherwise the point O is selected again, and finally each connected point O in different regions is marked to form a final segmentation image.

5. The bench line extraction method of claim 1, wherein: The concave hull algorithm extracts the profile of the segmented feature points in step S4, which comprises the following specific steps: The center point R of the discrete points in the separated image is calculated, an initial boundary point A closest to the point R is found, and is stored in a boundary point set, a point X is found to make the included angle of the vector RA counterclockwise rotating to the vector RX minimum, the included angle of the vector RA and any vector RX is calculated through the included angle formula, and the iteration is stopped until the selected point X coincides with the initial boundary point A, and the next step is performed.

6. The bench line extraction method of open pit mine according to claim 1, characterized in that: The step S4 of simulating the contour feature points by the Bezier curve comprises the following specific steps: According to the required selection of the corresponding number of data points during simulation, the curve passes through the start point and the end point inside the data points, and is tangent to the start point and the end point at the first and last sides of the feature polygon, and the intermediate point pulls the curve to itself; According to the required selection of the corresponding number of control points during simulation, when the number of control points is 1, the curve is a first-order Bezier curve, when the number of control points is 2, the curve is a second-order Bezier curve, and when the number of control points is 3, the curve is a third-order Bezier curve.

7. The bench line extraction method of open pit mine according to claim 1, characterized in that: The extraction method further comprises point cloud volume calculation, and the point cloud volume calculation comprises the following specific steps: (1) constructing two-period point cloud models based on step S1; (2) calculating the distance between the two-period point clouds to extract non-overlapping point clouds; (3) comparing with a threshold value, when the distance is less than the threshold value, the point cloud is an overlapping point cloud, and is removed, and when the distance is greater than the threshold value, the point cloud is a non-overlapping point cloud, and the next step is performed: (4) based on step S4, the non-overlapping point clouds are segmented by a region growing algorithm, point clouds with large number differences between the two periods are removed, each class of point clouds after processing is traversed, a slice plane and a slice step are determined to slice the point clouds, and the point clouds in the slice range are projected onto the slice plane; (5) based on step S4, the projected point cloud contour is extracted by a convex hull algorithm, the point cloud contour area is calculated by using Green's formula, the contour area is multiplied by the step to obtain the slice area, and each slice volume is added to obtain the point cloud volume of the part.

8. The bench line extraction method of claim 7, wherein: In step (5), before Green's formula is used, the projected point cloud contour needs to be sequentially investigated in terms of curve form, direction and connection mode, when the curve is a closed curve, the direction is a positive direction, and the connection mode is partial derivative continuity on a closed region, the subsequent calculation is performed by Green's formula, and if one of the conditions is not met, the point cloud is determined as an overlapping point cloud and is removed.

Images