Research On Registration Method Based On Multi-line Laser Point Cloud Data

With the upgrading of 3D measurement equipment and the rapid development of computer data processing technology,3D scanning technology can quickly and accurately obtain the point cloud data information from the object surface.The 3D scanning technology based on multi-line structured light is widely used in the collection of point cloud data because of its wide application environment,convenient equipment installation and easy commercial development.The initial point cloud data needs to be processed by point cloud data processing technology to extract various feature information from the point cloud data to meet the application requirements of various fields such as artificial intelligence,reverse engineering,and industrial automation.In this paper,the dense registration technology of multi-line structured light sparse point cloud,the coarse registration method of multi-line laser point cloud based on boundary centroid and the registration technology of multi-line structured light sparse point cloud based on non-coding marker points are studied.The specific contents are as follows:1.Aiming at the problem that the distance between light stripes of multi-line structured light is large and the geometric characteristics of point cloud data are not obvious,a denssification method for multi-line structured light point cloud is proposed.Extract the point cloud data which on the same surface of the measured object,use the random sample consensus algorithm or least square method to fit the equation of the object surface,and supplement the point cloud data between light stripes according to the fitting equation to make the original sparse point cloud densification,and then use the constraints between the surfaces of the measured object to eliminate redundant point cloud data.Experimental results show that this method can effectively fill in the missing point cloud data between light stripes and improve the accuracy of 3D point cloud reconstruction.2.Aiming at the problems of time-consuming and error matching in traditional four points registration of dense point cloud data,a coarse registration method based on boundary centroid is proposed.By extracting the boundary of point cloud data,it not only retains the appearance features of point cloud,but also reduces the size of point cloud data,which improves the speed of rough registration;in order to speed up the extraction speed of boundary points,the k-d tree algorithm is used to search for k nearest neighbor points;by registering the centroid of the boundary points of the point cloud,the initial distance of the point cloud is reduced and the overlap is increased,which ensures the accuracy of the coarse registration.Experiments show that the algorithm can effectively register the dense multi-line laser point cloud,and improve the speed and accuracy of coarse registration.3.In order to solve the problem that the point cloud data based on multi-line structured light is sparsely distributed,and it is difficult to extract feature descriptors through topological relations for point cloud registration,a multi-line structured light point cloud registration method based on non-coding marker points is proposed.Two landmarks are pasted on the measured object,and the landmarks under the same view angle are connected as vectors.The vector center is translated to the coordinate origin and the translation matrix is calculated.The rotation matrix between the landmarks under different views is calculated by Rodriguez formula.The transformation matrix of sparse point cloud registration is obtained by translation and rotation matrix.Experiments show that this method can quickly and accurately complete the registration task of sparse point cloud under multi-view.