Research On 3D Point Cloud Intelligent Evolutionary Registration And Surface Texture Mapping Method

With the wide application of 3D reconstruction in various fields of social life such as industry,medicine,aerospace and so on,the research on 3D reconstruction has become more and more in-depth.Among them,the 3D point cloud registration technology can unify the point cloud model obtained from different viewpoints into a coordinate system,thereby obtaining a complete 3D geometric model of the reconstructed object.But it is not enough,texture mapping must be performed to obtain a 3D scene or model which have a strong sense of reality,rich in detail and an immersive experience.Therefore,this paper analyzes and studies the 3D point cloud registration technology and texture mapping method.The main work and innovation are as follows:1、The point cloud registration method based on evolutionary algorithm is deeply studied.Firstly,the related basic theory of point cloud registration is introduced.Then,three widely used evolutionary algorithms are introduced and analyzed in detail.The algorithm structure is given and the advantages and disadvantages are discussed.Finally,the meaning of the evolution registration algorithm is clarified,the feasibility of the evolutionary algorithm to solve the registration problem is analyzed and the framework is given.2、This paper proposes a registration method based on improved backtracking optimization search algorithm.Firstly,for the unbalance problem of the backtracking search optimization algorithm with strong global exploration ability and relatively weak local development ability,three improved strategies are proposed: the mutation scale factor based on t distribution,the improvement of the structure of crossover equation and the use of evolutionary selection mechanism.Then the numerical optimization comparison experiments select 15 test functions for simulation.The results show that the proposed algorithm has obvious advantages in terms of convergence rate and search accuracy.Finally,this paper conduct registration experiments with point cloud library model and real shot model by using the algorithm.The results show that the algorithm can complete the registration task quickly and efficiently.3、This paper studies a high resolution texture mapping method based on rotating axis calibration method.The method combines the transformation relation of the camera basic coordinate system with the rotation axis calibration method,so that it does not need complex operations such as feature point extraction and matching,and only needs to perform real-time calibration on the left and right cameras when the main texture image is captured,and then can calculate the coordinate mapping relation between the other texture images and 3D point cloud.Then this paper conducts an experiment under laboratory conditions to verify this method.The coordinate mapping relation between high resolution texture images and 3D point cloud are successfully found and the texture map is carried out.A 3D reconstruction model with high resolution texture is obtained.The feasibility of the method is verified.