| In recent years,the precision measurement technology of aerospace parts has developed rapidly,but in the precise measurement of objects of large aerospace parts,when using a 3D scanning device for a single measurement,only the high-precision 3D topography information of the local surface of the target object can be obtained.Therefore,in order to more intelligently and comprehensively perform high-precision 3D scanning of the target object covering the complete shape,this paper uses the robot arm for trajectory planning,so that the robot arm carries the structured light 3D scanning equipment to realize automatic 3D scanning measurement of the measured object.After 3D scanning and measurement to obtain point cloud data from different perspectives,in order to obtain the 3D shape and position information of the measured object in a wider scene in a unified coordinate system,the point cloud of high-resolution point clouds from different perspectives of the measured object is further developed.Research on splicing methods.The main research contents of this paper are as follows:Firstly,aiming at the 3D scanning requirements covering the shape of large target objects,the 3D scanning trajectory planning based on the robotic arm is carried out.Design and implement a ROS-based scanning measurement simulation interface platform to set measurement points and path planning for the scanning trajectory of the robotic arm carrying the 3D scanning equipment to ensure intelligent and efficient data collection in real scenarios.The scanning measurement simulation interface outputs the scanning path data between the measurement points of multiple perspectives to the physical robot arm control cabinet,and then controls the robot arm to carry the 3D scanning equipment to collect data according to the established trajectory,so as to obtain the high-resolution point clouds of the measured target object.In the conventional 3D topography measurement scene,the measured object is selected as a standard workpiece.In view of the symmetrical structure of the standard workpiece and the flat appearance characteristics of the surface,this paper adopts the point cloud splicing method based on marked points to obtain the overall measurement 3D topography and location information of the measured workpiece in the reference coordinate system.In the process of point cloud splicing based on marked points,this paper improves the rough matching method of marked points based on distance matrix,and proposes a precise matching method of marked points based on congruent triangle constraints,the coarse matching pair of markers is purified to obtain the precise matching set of markers.After calculating the point cloud transformation matrix of adjacent viewpoints based on the precise matching set,the global registration algorithm of point cloud based on marker points is used to realize the overall 3D topography data of the target workpiece in the reference coordinate system.For the point cloud splicing of target objects with relatively complex shapes,since the splicing method based on marked points extracts the center features of the marked contours pasted on the surface of the object,therefore,it is more suitable to use the method of directly extracting the surface features of the point cloud for automatic stitching of the point cloud with higher accuracy.In this paper,a feature point detection algorithm integrating multi-features is proposed for the characteristics of high-resolution point clouds in the process of ”coarse-to-fine” pairwise splicing.On the fine registration method,an improved ICP algorithm based on outlier removal is proposed to improve the pairwise stitching accuracy.Considering the cumulative error of pairwise splicing,the global ICP algorithm is used to optimize the pairwise splicing results globally,and finally achieve accurate splicing of high-resolution point clouds without auxiliary information,thus,its overall three-dimensional shape and position information in the reference coordinate system are obtained. |
