摘要
针对目前轮廓特征识别算法存在识别结果易受噪声数据干扰、需对每一图形轮廓测量数据进行矢量化等问题,对二维截面轮廓特征重建方法进行研究,提出采用分步方式进行截面轮廓拟合的方法.首先,用平均增量法将测量数据进行直线段分段,并用最小二乘法对分段数据进行拟合,再把拟合结果存入链表内;然后,对于直线段链表内的相邻3条直线段,若相邻2条线段形成的半径满足给定阈值,则进行直线段圆弧拟合处理;最后,搜索拟合的圆弧段,如果存在2个或2个以上的连续圆弧,则将其重构为B样条曲线,得到二维截面轮廓特征的重构图形.该方法能有效减少拟合特征单元的数目,提高拟合精度.
With the problems that the recognition methods on profile feature are liable to be affected by noise data and the data of every profile of figures has to be transformed into vector data, the reconstruction method on profile features of 2D section is studied and a new method is proposed to implement 2D section profile fitting step by step. Firstly, the measurement data is divided into line points with the average incremental method, all the line points are fitted with Least-Squares Fitting （LSF）, and the results are saved into a list table ; secondly, with three continuous lines in the fist table, if the radius created by two of them satisfies the given threshold, the lines are fitted to arcs ; at last, if there are two or more continuous arcs by searching all arcs, these arcs are reconstructed into B-splines and the reconstructed graphics of 2D section profile feature are obtained. The method can effectively decrease the number of fitting feature elements number and improves fitting precision.
出处
《计算机辅助工程》
2008年第3期18-22,共5页
Computer Aided Engineering
