| With the development of point cloud equipment,people can quickly and easily access the actual scene of a large number of point cloud data,and promote the development of reverse engineering technology.In the process of 3D reconstruction,in order to obtain a wide range of scenes of cloud,from different angles to get the scene point cloud data,so point cloud stitching and reconstruction technology is particularly important.This paper s tudies the splicing and reconstruction algorithm of point cloud data processing and related problems.There are some problems such as time-consuming and poor splicing in point cloud stitching.In this paper,we study the existing ICP algorithm and the method of obtaining transform matrix by LM and SVD.By using the orthogonal mapping,the mapping relation between the 3D point cloud and the 2D image is obtained,and the matching point pair is obtained in the 2D image.A method of combining the LM and SVD method is proposed to obtain the transformation matrix.Finally,the experimental results of several mosaic algorithms are analyzed and compared,and the improved algorithm proposed in this paper is proved to reduce the stitching error while greatly reducing the splicing time.Aiming at the problems such as poor anti-noise,time-consuming,rough and existing holes in point cloud reconstruction.In this paper,the poisson algorithm and the greedy projection triangulation algorithm are studied.Then,based on th e greedy projection triangulation algorithm,firstly,an improved weighted moving least squares method is proposed to carry out the normal estimation.The reconstruction effect is smooth and has anti-noise.Secondly,improved preprocessing method,the data resampling on a smooth basis makes the reconstruction effect smoother and refine the hole.Finally,for the long time to reconstrection,the use of octree instead of KD tree,the realization of the reconstruction of the aging optimization without affecting the reconstruction of the foundation. |
PDF Full Text Request