Technology Of Point Cloud Stitching Based On SIFT And Random Dots

With the development of computer,the benefits of using machine vision to measure objects become more and more obvious.Using binocular vision to detect has the characteristics of high accuracy and strong robustness,which meets the needs of high accuracy for measuring objects.Because of this characteristic,the detection technology has become one of the fast developing technologies.Because the measured object may be a curved surface or a large measurement size,it is impossible to obtain complete parameter information through one shot due to the camera field of view under the condition of ensuring a certain resolution.So it is an important part of measuring curved objects how to make a better joint of point and cloud,and how to unify the parameters of objects that are photographed many times.This paper mainly studies the point cloud stitching technology in binocular stereo vision detection system.Based on the analysis of the principle of surface splicing technology and binocular vision detection method,a binocular stereo vision surface point cloud splicing system based on random points is built.The number of feature points is increased by adding gray information based on random points to low texture and smooth surfaces.The increase of the number of feature points plays a positive role in the coarse splicing of point cloud stitching.By using the octree to voxel the discrete point cloud from binocular vision system,the voxel which can be compared with the two-dimensional image pixel can be obtained.The SIFT operator(scale invariant feature transform)is used to extract feature points from the point cloud with random point information.Rough matching of surface point cloud is carried out according to the extracted feature points.Tricp(trimed iterative closure point),an optimized method for traditional ICP(iterative closure point),is used to further match the coarse matching point cloud.Thus,the problem of low repetition region matching caused by curved objects can be solved.After the point cloud stitching of binocular stereo vision,the results of many experiments show that the variance of the spatial distance of the specific code points on the spliced object is small,and the experimental data is stable and reliable,which shows that the robustness of the splicing algorithm proposed in this paper is ideal,while the theoretical analysis conclusion relied on by the practical method is correct and practical.