基于边缘检测的曲面工件定位预处理方法  

Edge detection based pre-process for work-piece surface localization

作　　者：王冬[1] 周凯[1] 

机构地区：[1]清华大学机械工程系,北京100084

出　　处：《中国科技论文》2015年第22期2596-2600,共5页China Sciencepaper

基　　金：国家自然科学基金资助项目(50775126);高等学校博士学科点专项科研基金资助项目(20110002110079)

摘　　要：利用弹性工件刚性化加工技术,可使弱刚度曲面工件的定位问题转化为点云的刚体配准问题。迭代最近点(ICP)算法是点云刚体配准最常用的方法,但当2个点云偏移较大时可能会陷入局部极小。工件合理的初始位置可由人工操作来保证,但这种方式严重依赖于操作者的经验且精度有限,特别是对于毛坯件等缺少精确定位信息的工件就更加困难。本文提出一种基于边缘检测的预处理算法,利用少量边缘点计算最优的调整方向,将工件调整到合适的位置,以确保后续ICP方法的有效性。仿真结果表明,该方法能有效解决缺失点对ICP性能的影响,改善传感器覆盖率指标。Localization of weakly rigid work-piece is converted to rigid registration of two-point clouds by using flexible tooling system.Iterative closest point（ICP）is widely used in rigid registration.But its performance decreases apparently when the relative offset of two-point clouds is large,because of possible local minimum.Good initial position can be guaranteed through manual operation,which relies on operator＇s experience.And it is much harder to precisely localize a semi-finished work-piece due to the error and lack of information.In this paper,a edge detection based pre-process algorithm is proposed.By using a small amount of edge points,a transform direction is calculated to lead the work-piece to the better position,which guarantees the ICP prerequisite.The results show that the combined method could reduce the negative impact of missing parts to ICP,and gain better sensor array coverage.

关 键 词：曲面定位 边缘检测 点云配准 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]