Research On 3D Rock Segmentation And Grab Selection Method

As an important part of deep space exploration,asteroid detection has attracted much attention.The detection of asteroids through asteroids can not only promote the development of aerospace technology,but also further enhance people’s knowledge and understanding of the universe.By sampling the asteroid and further studying its samples,not only can the physical and chemical properties of the asteroid itself be analyzed,but also the natural resources on the asteroid can be found and explored,and the current status of the universe can be further studied,To analyze the evolution of the universe and increase the cognition and understanding of the universe.This paper analyzes the advantages and disadvantages of the sampling methods of each spacecraft in the existing asteroid sampling task.Because the existing sampling methods are random,the research value of the samples is reduced,so a visual method is needed to the method of sampling the sampled samples before sampling,so that the detector can obtain better samples.The asteroid rock area has the following characteristics:(1)The rock texture is complex and irregular,the same rock contains multiple textures,and the same texture appears on multiple rocks,and the rock outline cannot be effectively obtained through the two-dimensional image gray value information;(2)The shapes of rocks are different and irregular,there are no obvious characteristics of curvature and normal vector,and the curvature and normal vector change irregularly.The three-dimensional information such as curvature and normal vector cannot effectively divide the rock area.In view of the particularity of the existence of the rock area,this paper proposes a method of rock segmentation using the three-dimensional information of the rock.(1)Establish a binocular vision system,perform camera calibration,and complete the settlement of the camera’s internal and external parameters.(2)Use the camera to shoot the rock area,get the image information,and reconstruct the 3D point cloud of the rock area.The method of three-dimensional point cloud reconstruction adopts the vertical line trajectory method.According to the particularity of the asteroid detection task,the high-resolution three-dimensional point cloud reconstruction of the entire area takes too long,so it is based on the vertical line trajectory method.Improvement,using prior knowledge as an elevation reference,firstly perform low-resolution 3D point cloud reconstruction on the entire area,and then select a local area among them to perform high-resolution 3D point cloud reconstruction.This method can effectively reduce plumbness.The number of elevations selected in the line trajectory method improves the speed of 3D point cloud reconstruction.(3)Use the moving cube algorithm.The surface reconstruction is completed by calculating the relationship between the voxel vertex and the isosurface.(4)Divide the rocks.As the basis of rock division,the concavity and convexity are selected,and the concavity and convexity of two adjacent planes are selected by LCCP method,and all planes are traversed by regional growth,and the adjacent planes that constitute the convex structure are clustered to complete the rock segmentation.Experimental results show that the method proposed in this paper can effectively complete rock segmentation and grab selection in the actual rock area.