Curve Fitting Based On Point Cloud

The titles of paper is from the national natural science fund'Virtual three -dimensional model reconstruction based on heterologous RS data'and'Researching on three-dimensional model reconstruction based on terrestrial laser scanning point cloud of complex structures systems and close-range image'.The main work of this paper is boundary curve fitting from 3D laser scanner.The purpose of curve fitting control surface reconstruction of three-dimensional data,in order to realize stereo display of three-dimensional objects.This paper mainly finished work:In this paper,the original extraction,especially the methods that are common used in recent years are in-depth studied,then boundary extraction of point cloud based on image features are proposed.This paper is fully consider features of point cloud and image data,combining edge information clarity for image data and point cloud accuracy,and improve the accuracy of boundary extraction of point cloud.Curve fitting based on the boundary point cloud that are extracted.we construct'Tree Chain of Dominant Point'to get the main characteristics of curve and secondary feature points.Then,the feature points of extracted adopt cumulative chord length method to realize parameterization.Finally curve fitting is completed by cubic B-spline algorithm.In the same time, the curve of nodes is optimized. Purpose is to guarantee the curve fitting precision under the requirement of the premise, with minimal points to express curve.Experimental results demonstrate the effectiveness of the curve fitting algorithm.