Research On Registration Technology Optimization Of Scattered Point Cloud

Reverse engineering technology is an important assist technology in product design.It plays an important role in the development of science and technology in China.With the use of reverse engineering technology in all walks of life,it greatly reduces the variety of product design cycle.In order to get the complete point cloud model accurately,multiple scan models are needed to register all the point clouds to form a complete point cloud model.This paper proposes new solutions and improvements for the existing problems of rigid registration methods.The correctness and effectiveness of the proposed scheme were verified by comparative experiments.The main work of this article includes:(1)In this paper,Point cloud registration based on entropy criterion genetic algorithm is proposed.This method has strong robustness,and it has good effect for only a few overlapped point cloud data and noisy point cloud data.The method uses the new method for evaluating the location of point cloud---spatial distribution entropy(SDE).The entropy of spatial distribution is used to evaluate the relative position of two clouds based on the concept of information entropy and the relationship between point cloud spatial coordinates and density.The spatial distribution entropy is used as the optimization function of the point cloud registration,and the genetic algorithm is used as the optimization method.The combination of the two achieves the point cloud coarse registration.Experiments show that the spatial distribution entropy can greatly improve the spatial position efficiency compared with the mean squared error.The search algorithm of the genetic algorithm has a directionality to avoid global search and greatly improve the efficiency of registration.(2)In this paper,An improved ICP point cloud fine registration algorithm based on point cloud feature histogram is proposed.This method uses the point cloud feature histogram to have a good discrimination degree,and can accurately form the corresponding point pair when searching for corresponding points,so that the ICP algorithm converges faster and also improves the precision of point cloud registration.(3)In this paper,a universal registration algorithm is proposed.The algorithm adopts the first rough registration strategy for fine registration.Coarse registration uses spatial distribution entropy as optimization objective and combines genetic algorithm that is directional search algorithm.It can quickly search space optimal transformation,and is effective for a small number of overlapping point clouds and noisy point clouds.Fine registration improves the accuracy of registration by adding high-dimensional feature point histograms when coarse registration provides a good initial position.The perfect combina-tion of the two algorithms makes use of their own advantages to make the registration algorithm universal and accurate.