Research On The Algorithms Of Point Cloud Pre-Processing And Feature Corner Detection

Reverse Engineering is one of the methods for digital product development. It shortens exploitation period, improves precision and is an important way to create and develop a new product by digesting and absorbing advanced techniques. The measurement of shape and scale is one of basic methods which are used to rebuild the world by mankind.Presently, research on the pivotal technology of RE, which includes data acquiring, data dealing, feature extraction and model reconstruction. Data acquiring is the basis of RE; Data dealing, which is the crucial tache of RE, will affect the result of model reconstruction; Model reconstruction is the ending. On account of that, this thesis will focus on data dealing and feature extraction.The main contribution of this paper is that, creative idea is proposed especially in point cloud sampling, the extraction of boundary feature curve and corner decetion based on analyzing and summarizing the current algorithm.The study for preprocessing of data points in this paper is mainly concentrated on the noise error reduction of points cloud, data points smoothing, data points simplify and data points egmentation. First of all, we introduces the techniques of the noise error reduction of point cloud data and data points smoothing and propose an algorithm based on image to reduce noise in point clouds. Then we introduce the techniques on the existing algorithms of building the topological relationship of scattered points. At last we introduces the techniques on the existing algorithms of data points Simplify, this paper also realizes a data reduction method on the basis of the value of chord.This paper researches on the existing algorithms of feature extraction,then base on the characteristic of points cloud with laser stripe structured light ,we propose an feature extraction algorithm of 3D plane points based on Mathematical Morphology.Further more, two kinds of methods about feature extraction of 3D points cloud is presented. One kind of feature extraction is based on the curvature value of the point; another kind is based on the distributing of K-Nearest Points.Through research on former comer detection techniques, an adaptive comer detection algorithm according to the boundary is proposed in this paper. The principle is described in detail. These methods are proved to be effective by some examples.