基于曲率及误差分布的点云逼近曲线优化算法  

Optimization algorithm for approximate curve of point-clouds base on curvature distribution and error distribution

作　　者：张蓉生[1] 张伟华 马凌腾[1] 米紫昊[1] 

机构地区：[1]河海大学能源与电气学院,江苏南京210098 [2]摩托罗拉(中国)成都分公司,四川成都611731

出　　处：《计算机集成制造系统》2011年第5期1017-1023,共7页Computer Integrated Manufacturing Systems

摘　　要：为提高利用形状信息优化点云逼近曲线的准确性,提出了一种基于曲率及误差分布的点云逼近曲线优化算法。该算法首先快速地创建满足误差限的非最优样条曲线,据此获得较为准确的曲率分布作为形状信息;然后,以大于所给曲率阈值的角点作为种子点,构建出点云的三次序号参变量样条逼近曲线。当逼近曲线误差不满足误差限时,以误差的分布情形确定型值点的添加位置,使得每一次型值点添加后最大误差均减小。通过数条点云曲线实验表明,该算法显著地减少了逼近曲线的控制点数目。To improve accuracy of shape information so as to optimize the approximate curve of point-clouds, an optimization algorithm was proposed based on curvature distribution and error distribution. Firstly, a non-optimal Bspline curve within the error limit was rapidly established to obtain more accurate curvature distribution. And then, the corner points which were greater than the curvature threshold were chosen as the seed points. The cubic splint approximate curve with sequence number parameters was established by this seed points. If the error of approximate curve was greater than the error limit, the added position of a new dominant point was determined based on the error distribution so as to decrease the maximum error after adding every time. Experiments showed that the control points number of approximate curves was remarkably decreased.

关 键 词：点云 优化 曲率分布 误差分布 逼近曲线 控制点 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]