基于图像的轮廓度测量与评定  

Measurement and evaluation for profile tolerance based on images

作　　者：王仲[1] 操晶晶[1] 张立昆[1] 张进[1] 

机构地区：[1]天津大学精密测试技术及仪器国家重点实验室,天津300072

出　　处：《光学精密工程》2009年第2期395-401,共7页Optics and Precision Engineering

基　　金：天津市自然科学基金资助项目(No.043612111)

摘　　要：针对微小型平面零件给出了一种基于图像的轮廓度测量与评定方法。包括实现轮廓度测量的理论轮廓数据建模方法、对应理论点的扩展极角定界搜索方法以及满足最小条件的最小极偏差轮廓度误差评定方法。通过实例将最小极偏差轮廓度评定方法（LDM）与传统的最小二乘轮廓度评定方法（LSM）进行了分析和比较。结果证明，对于不同实际零件，最小极偏差评定方法较最小二乘评定方法收敛于最小值的迭代次数减少了0～3次，精度优化了4％～12．6％，可以满足微型零件高精度、高效率轮廓度检测的要求。A profile measurement and evaluation method for the micro-parts based on the images is presented. This method includes the method of profile data modeling in theory, bound searching method of expansion polar angle corresponding theoretical points and evaluation method of least-deviation profile error. The evaluation method of least deviation profile error and classical least -square profile error are analyzed and compared based on several examples. The experiments on actual image show that the iterative times of the least deviation method are 0-3 fewer and the accuracy are 4%- 12% higher than that of least-square method, which can meet the requirements of high accuracy and fast profile tolerance measurement of micro-parts.

关 键 词：微型零件 轮廓度误差 图像处理 最小极偏差 最小二乘 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]