Method for calculating boundary and volume of material pile based on shielded material pile three-dimensional point cloud

Abstract

The invention provides a method for calculating the boundary and the volume of a material pile based on sheltered material pile three-dimensional point cloud. The method comprises the following steps: step 1, collecting three-dimensional point cloud data of the surface of a whole stock bin in real time; 2, converting the original point cloud data into a specified world coordinate system, and performing point cloud filtering and point cloud segmentation on the three-dimensional point cloud data in the coordinate system; step 3, extracting each material pile point cloud in the three-dimensional point cloud data by using an improved region growing algorithm; 4, calculating the boundaries of the material piles through a concave packet and KD tree method, and complementing the point clouds at the shielded positions of the point clouds of all the material piles; and 5, calculating the volume of each material pile through a slicing method by taking the plane where the boundary of the material pile is located as a bottom surface. According to the method, point cloud supplement is carried out on the shielded material pile point cloud, and the problem that traditional material pile volume calculation can only be carried out through the complete material pile point cloud is solved.

Description

Technical Field

[0001] The present application relates to the field of point cloud data, and in particular to a method for computing the boundary and volume of a material pile based on an occluded three-dimensional point cloud of a material pile. Background Art

[0002] In the production process of concrete mixing plants, various grades of stone powder, machine-made sand, tailings, etc. are required as raw materials for mixed concrete. There are multiple piles of materials in the material yard, and the total volume is often large. Real-time calculation of the boundaries and volumes of various raw materials is of great significance to the safety and continuity of the entire concrete mixing plant production process. Since the material piles are irregular in shape when they are stacked, it is difficult to directly measure the volume of the material piles.

[0003] 3D laser point cloud scanning technology has been widely used in the fields of construction, production and inspection in recent years due to its high efficiency, high precision and strong data operability. This technology uses a 3D laser scanner to perform non-contact scanning on the target object, obtain the spatial coordinates of each point on its surface, and thus construct a 3D point cloud model. This technology can not only realize the digital modeling of the whole, part and details of the target object, but also can be applied to the volume detection of the stockpile, quickly generate a 3D point cloud model, and obtain the volume and quality information of the raw materials, which greatly saves manpower and material resources. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some embodiments. The purpose of the abstract and the invention title in this section and the present application is not to be ambiguous, and such simplification or omission cannot be used to limit the scope of the present invention.

[0005] The object of the present invention is to provide a method for computing the boundary and volume of a pile of materials based on occluded three-dimensional point clouds, so as to complete the real-time calculation of the boundary and volume of the pile of materials.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: step one, real-time collection of three-dimensional point cloud data of the entire silo surface; step two, converting the original point cloud data to a specified world coordinate system, and performing point cloud filtering and point cloud segmentation on the three-dimensional point cloud data in the coordinate system; step three, using an improved regional growing algorithm to extract point clouds of each material pile in the three-dimensional point cloud data; step four, calculating the boundary of the material pile and completing the point clouds of the obscured positions of each material pile point cloud by the concave hull and KD tree methods; step five, taking the plane where the boundary of the material pile is located as the bottom surface, calculating the volume of each material pile by the slicing method.

[0007] Furthermore, step one includes: the present invention collects three-dimensional point cloud data of the entire silo surface in real time through a laser scanner. The laser scanner is arranged just above the front windshield on the top of the loader cab, and the three-dimensional point cloud data can be collected while the loader is loading.

[0008] Furthermore, step two includes: after transforming the original three-dimensional point cloud data into a unified coordinate system through rotation and translation, pre-processing the three-dimensional point cloud data, based on the point cloud intensity characteristics and the distance between points, taking the material pile point cloud intensity as a benchmark, traversing the three-dimensional point cloud data to complete the point cloud noise removal; by calculating the distance between points in the neighborhood of each three-dimensional point, and based on the mean and standard deviation of these distances, completing the outlier removal; and then segmenting the filtered three-dimensional point cloud data by pre-setting coordinates and random sampling consistency, and retaining the three-dimensional point cloud data corresponding to each material pile.

[0009] Furthermore, step three includes: using an improved region growing algorithm to extract point clouds of each material pile in the three-dimensional point cloud data, refining the three-dimensional point cloud data of each material pile without the ground, removing other noise points outside the material pile point cloud, obtaining three-dimensional point cloud data of each material pile that meets the accuracy, and improving the robustness of the algorithm.

[0010] Furthermore, step four includes: calculating the boundary of the material pile by means of the concave hull and KD tree methods, projecting the material pile point cloud data that meets the accuracy requirements onto the XOY plane in the world coordinate system, extracting the boundary of the plane point cloud using the concave hull algorithm, and then restoring the plane point cloud boundary points to the boundary of the material pile three-dimensional point cloud by means of the KD tree algorithm, setting a height threshold to filter out the boundary points of the three-dimensional point cloud of the material pile; and adding a preset number of points at a fixed spacing in the elevation direction between the plane point cloud boundary points and the corresponding three-dimensional point cloud boundary points of the material pile.

[0011] Furthermore, step five includes: calculating the volume of each material pile by a slicing method, adaptively slicing according to a set number of slices by calculating the difference in height between the highest and lowest points of the three-dimensional point cloud data of the material pile, stratifying the three-dimensional point cloud data of the material pile at equal intervals in the vertical direction, merging the slices with too few points in the three-dimensional point cloud data of the material pile with the previous layer, recording the final number of layers, using a convex hull algorithm to calculate the boundary point set of each slice, connecting adjacent boundary points and each boundary point to the centroid of the boundary point respectively to divide the slice into multiple triangles, calculating the areas of all triangles and summing them as the area of ​​the slice, using the trapezoidal prism volume formula to calculate the volume of the point cloud between two slices respectively, and summing the volumes between all slices to obtain the real-time volume of each material pile. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0013] Figure 1 It is a flow chart of the present invention;

[0014] Figure 2 The present invention uses a concave hull algorithm to extract a schematic diagram of the stockpile boundary;

[0015] Figure 3 A schematic diagram of adding a point cloud between a plane boundary point and a corresponding three-dimensional point cloud boundary point of a pile of materials according to the present invention;

[0016] Figure 4 This is a schematic diagram of a middle-layer slice of a three-dimensional point cloud of a pile of materials according to the present invention. DETAILED DESCRIPTION

[0017] Reference Figure 1 A method for calculating the boundary and volume of a material pile based on an occluded three-dimensional point cloud of a material pile comprises the following steps: step one, real-time acquisition of three-dimensional point cloud data of the entire material silo surface; step two, converting the original point cloud data to a specified world coordinate system, and performing point cloud filtering and point cloud segmentation on the three-dimensional point cloud data in the coordinate system; step three, using an improved regional growing algorithm to extract each material pile point cloud in the three-dimensional point cloud data; step four, calculating the boundary of the material pile and completing the point cloud of each material pile point cloud that is occluded by the concave hull and KD tree method; step five, taking the plane where the boundary of the material pile is located as the bottom surface, calculating the volume of each material pile by a slicing method.

[0018] Step one includes: the present invention collects three-dimensional point cloud data of the entire silo surface in real time through a laser scanner. The laser scanner is arranged just above the front windshield on the top of the loader cab, and the three-dimensional point cloud data can be collected while the loader is loading.

[0019] Step two includes: after transforming the original three-dimensional point cloud data into a unified coordinate system through rotation and translation, pre-processing the three-dimensional point cloud data, based on the point cloud intensity characteristics and the distance between points, taking the material pile point cloud intensity as a benchmark, traversing the three-dimensional point cloud data to complete the point cloud noise removal; by calculating the distance within the neighborhood of each three-dimensional point, and based on the mean and standard deviation of these distances, completing the outlier removal; and then segmenting the filtered three-dimensional point cloud data by pre-setting coordinates and random sampling consistency, and retaining the three-dimensional point cloud data corresponding to each material pile.

[0020] Step three includes: using an improved region growing algorithm to extract each material pile point cloud in the three-dimensional point cloud data, refining the three-dimensional point cloud data of each material pile without the ground, removing the remaining clutter outside the material pile point cloud, and improving the robustness of the algorithm.

[0021] Based on the similarity formula between the three-dimensional point cloud data points of each stockpile without the ground

[0022] S(Pi, Pj)=||Pi-Pj||<T

[0023] Filter out the three-dimensional point cloud data of each stockpile that meets the accuracy requirements, where Pi and Pj represent the feature vectors of the three-dimensional points, and T represents the feature vector angle threshold.

[0024] Reference Figure 2 , Step 4 includes: calculating the boundary of the stockpile by using the concave hull and KD tree methods, projecting the three-dimensional point cloud data of each stockpile that meets the accuracy onto the XOY plane, and taking the average distance between point clouds as the value of the minimum inner angle Alpha allowed for the concave hull boundary. The larger the Alpha value, the larger the minimum inner angle allowed, the smoother the concave hull, and the closer it is to the convex hull; the smaller the Alpha value, the smaller the minimum inner angle allowed, the more concave the concave hull, and the closer it is to the actual shape of the point cloud; using the KD tree algorithm, recursively dividing the three-dimensional point cloud data set into smaller subsets, making it more efficient to restore the spatial plane boundary to the three-dimensional boundary, and setting a height threshold to filter out the boundary points of the three-dimensional point cloud of the stockpile.

[0025] Reference Figure 3 ,Step 4 includes: adding a preset number of points between the plane point cloud boundary points of the ,corresponding three-dimensional point cloud boundary points of the pile ,according to the fixed spacing in the elevation direction, selecting the appropriate ,predetermined number, and inserting multiple sets of reasonable ,stock pile point cloud supplementary points to ensure that there is no volume omission problem at the ,smallest computational cost possible.

[0026] Reference Figure 4 ,Step 5 includes: taking the plane where the boundary of the pile is located as the bottom surface, calculating the volume of each pile by slicing method, adaptively slicing by calculating the height difference between the highest and lowest points of the three-dimensional point cloud data of the pile, stratifying the three-dimensional point cloud data of the pile at equal intervals in the vertical direction, merging the slices with less than 5 points in the three-dimensional point cloud data of the pile with the previous layer, recording the number of layers, using the convex hull algorithm to calculate the boundary point set of each slice, and calculating the area S of the slice. plan , divided into the sum of multiple triangle areas S1, S2..., each triangle is composed of adjacent boundary points and each boundary point is connected to the centroid of the boundary point. The calculation formula for each triangle is

[0027]

[0028] Where a, b, and c are the side lengths of the triangle, C is half of the perimeter of the triangle, and the three-dimensional point cloud volume of the pile between slices is calculated using the trapezoidal prism volume formula.

[0029] The present invention solves the problem that traditional material pile volume calculation requires a complete material pile point cloud to calculate by using a method for supplementing the obstructed material pile point cloud.

[0030] The above is only a preferred embodiment of the present invention. It should be noted that for ordinary persons in the art, various modifications can be made without departing from the scope of protection of the present invention. Although the above embodiments are described in detail, more other equivalent embodiments may be included without departing from the principle of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for calculating the boundary and volume of a pile of materials based on occlusion three-dimensional point cloud, characterized in that: The following steps are involved: Step 1: collect the 3D point cloud data of the entire silo surface in real time; Step 2: convert the original point cloud data into a specified world coordinate system, and perform point cloud filtering and point cloud segmentation on the three-dimensional point cloud data in the coordinate system; Step 3, using an improved region growing algorithm to extract each stockpile point cloud in the three-dimensional point cloud data; Step 4: Calculate the boundary of the pile and complete the point cloud of each pile point cloud where it is blocked by the concave hull and KD tree method; Step 5: Using the plane where the boundary of the pile is located as the bottom surface, calculate the volume of each pile by using the slicing method.

2. The method for computing the boundary and volume of a pile of materials based on occlusion-based three-dimensional point cloud of a pile of materials according to claim 1, characterized in that: In step 1, the three-dimensional point cloud data of the entire silo surface is collected in real time by a laser scanner, and the laser scanner is arranged just above the front windshield on the top of the loader cab.

3. The method for computing the boundary and volume of a pile of materials based on occlusion-based three-dimensional point cloud of a pile of materials according to claim 1, characterized in that: In the step 1, the original point cloud data is converted to a specified world coordinate system, specifically: the silo entrance is set as the zero point of the world coordinate system, the west is the positive direction of the X-axis, the south is the positive direction of the Y-axis, and the top is the positive direction of the Z-axis. The original point cloud is converted to the world coordinate system by rotating and translating.

4. The method for computing the boundary and volume of a pile of materials based on occlusion-based three-dimensional point cloud of a pile of materials according to claim 1, characterized in that: In the step 2, the three-dimensional point cloud data is subjected to point cloud filtering, specifically: based on the point cloud intensity characteristics and the distance between points, the three-dimensional point cloud data is traversed with the point cloud intensity of the pile as a benchmark to complete the point cloud noise removal; by calculating the distance between points in the neighborhood of each three-dimensional point, and based on the mean and standard deviation of these distances, outlier removal is completed.

5. The method for computing the boundary and volume of a pile of materials based on occlusion-based three-dimensional point cloud of a pile of materials according to claim 1, characterized in that: In the step 2, the three-dimensional point cloud data is segmented, specifically: the three-dimensional point cloud data after filtering is segmented by a preset coordinate interval, and the three-dimensional point cloud data corresponding to each material pile is retained respectively; the three-dimensional point cloud data of each material pile is plane segmented by a random sampling consistency algorithm, and the three-dimensional point cloud data of each material pile minus the ground is retained respectively.

6. The method for computing the boundary and volume of a pile of materials based on occlusion-based three-dimensional point cloud of a pile of materials according to claim 1, characterized in that: In the step three, an improved region growing algorithm is used to extract point clouds of each material pile in the three-dimensional point cloud data. Specifically, the three-dimensional point cloud data of each material pile without the ground is refined, and based on the improved region growing method, the point cloud data of each region is evaluated for robustness, and the point cloud data of each region is filtered according to the evaluation results, so as to obtain three-dimensional point cloud data of each material pile that meets the accuracy.

7. The method for computing the boundary and volume of a pile of materials based on occlusion-based three-dimensional point cloud of a pile of materials according to claim 6, characterized in that: The improved region growing method in step three is specifically: overloading the original region growing algorithm function, and changing the output of the function into a C++ container by traversing each region.

8. The method for computing the boundary and volume of a pile of materials based on occlusion-based three-dimensional point cloud of a pile of materials according to claim 1, characterized in that: In the step 4, the boundary of the material pile is calculated by the concave hull and KD tree method, specifically: the three-dimensional point cloud data of each material pile that meets the accuracy is projected onto the XOY plane in the world coordinate system, the boundary of the projected point cloud is extracted by the concave hull algorithm, the boundary points of the plane point cloud are restored to the boundary of the three-dimensional point cloud of the material pile by the KD tree algorithm, and the points below the preset height value are used as the boundary points of the three-dimensional point cloud of the material pile.

9. The method for computing the boundary and volume of a pile of materials based on occlusion-based three-dimensional point cloud of a pile of materials according to claim 1, characterized in that: The step 4 of completing the point clouds of the obstructed positions of each stockpile point cloud is specifically as follows: a preset number of points are added at a fixed interval in the elevation direction between the boundary points of the plane point cloud and the corresponding boundary points of the three-dimensional point cloud.

10. The method for computing the boundary and volume of a pile of materials based on occlusion-based three-dimensional point cloud of a pile of materials according to claim 1, characterized in that: In the step 5, the volume of each stockpile is calculated by a slicing method, specifically: adaptive slicing is performed by calculating the height difference between the highest and lowest points of the three-dimensional point cloud data of the stockpile, the three-dimensional point cloud data of the stockpile is layered at equal intervals in the vertical direction, the slices with less than 5 points in the three-dimensional point cloud data of the stockpile are merged with the previous layer, the number of layers is recorded, and the boundary point set of each slice is calculated using a convex hull algorithm, adjacent boundary points and each boundary point are connected to the centroid of the boundary point to form multiple triangles, the sum of the areas of all triangles is calculated as the area of ​​the slice, and the volume of the point cloud between two slices is calculated respectively, and the sum is the real-time volume of each stockpile.

Images