A Research On Preprocessing Algorithms Of Mass Point Cloud

Reverse Engineering (RE) is a new technology in CAD area, which brings a revolution to the traditional CAD techniques. Usually, the data that they get from scanners exists as a "Point Cloud". The amount of points in these point clouds rising rapidly when laser scanners are widely using. We must do some preprocess to these "Mass Point Cloud", otherwise they will bring many disadvantages to the next processes.Through studying other techniques, three new algorithms belong to the preprocessing of mass point clouds are put forward in this paper. They are used for "Registration", "Noise reducing" and "Simplify" respectively.First of all, we propose an algorithm based on image to reduce noise in point clouds. It's independent of the amount of points. Therefore, it has a quite high speed. In our experiment, it also has a good result of noise removing. This paper also introduces an algorithm based on the k-d tree. It doesn't have the same speed as the former one, but it has good performance on sparse point clouds.Then we describe an accurate and efficient algorithm of point cloud auto registration. This algorithm contains two steps: initial registration and precise registration. We use eigenvectors of point clouds to achieve initial registration. For precise registration, an improved "Iterative Closest Point"(ICP) algorithm based on feature points generated by curvature is introduced. The result of the algorithm is verified in the applications.This paper also proposes an algorithm used for point cloud simplify. That is a mixed sampling algorithm including random sample and curvature sample. It can remove many points while keeping the characteristic of point clouds.