Studies On The Processing Of 3D Laser Scan Data

3D laser scanner can quickly measure the models of product and a variety of mold. It have a wide range of applications in the digital design and manufacture, such as cars manufacture, sports equipment, furniture production, copy of antique and artwork, three-dimensional animation, virtual reality, medical rehabilitation, body measure, E-commerce and so on..Due to a variety of physical factors of the acquisition procedure, point sets always are prone to various kinds of undesirable noise which will have adverse impact on accuracy of point data, so it's necessary to remove the effects caused by noises. In order to get the whole information on the model surface, we have to measure the model from different views. But there are many redundant points exist in the overlap region of two adjacent multi-view data, and the method which just merges two adjacent multi-view data can't obtain ideal model, so it's necessary to find a good algorithm to registrant two multi-view data.In this paper, we focus on the research on denoising, smoothing and the method to merge the point sets get from multi-views, and main contribution include:Since the classical denoising and smoothing methods are poor at preserve the details of point-sets of carving crafts, a smoothing algorithm using B-Spline base on edge detection was proposed to solve the problem. At first, we extract the edge of objects on gray image derived from point-sets, then in the area without any edges, we put forward the B-Spline to smooth point cloud.A introduction to classical edge extraction algorithms on image. After analyzing and summarizing the fundamental theories, the disadvantages and advantages of different methods are induced. We choose the Laplace operator to help us get the edge of objects on image.In order to merge the point sets get from multi-view, we propose a method based on section line which can joint point sets according to their position in slice .It can raised the speed and accuracy of registration of multi-view.At original software platform, we add a function module which can remove the undesirable noise and merge point gets from multi-view, so that point sets can achieve better smoothness and accuracy of point cloud registration.