基于显微视觉的微零件在线检测与装配策略研究  

On-line detecting and assembling of micro parts based on microscope vision

作　　者：张娟[1] 张正涛[1] 徐德[1] 

机构地区：[1]中国科学院自动化研究所精密感知与控制研究中心,北京100190

出　　处：《高技术通讯》2013年第8期848-855,共8页Chinese High Technology Letters

基　　金：国家自然科学基金(61105036和61227804)资助项目

摘　　要：基于搭建的微装配实验平台,提出了一种基于三路显微视觉的粗精结合的微零件在线检测与装配策略,实现了对mm级复杂几何结构微零件的在线检测与装配。该策略通过粗调流程和精调流程分别在最小放大倍数和较高放大倍数下实现对微零件相对位姿的检测与调整,从而实现微零件位姿对准与装配。基于图像特征区域跟踪,提出了一种自动变倍算法,在保证微零件特征区域位于视野范围内的同时,实现了对显微视觉系统的自动变倍。基于对微零件成像特征的分析,提出了基于图像特征的微零件相对位姿检测算法,实现了对微零件相对位姿的在线检测。实验结果表明,姿态检测误差小于0.5°,位置检测误差小于5μm。实验结果验证了所提算法的有效性。Based on an experimental platform constructed for micro-assembly,a strategy for online detection and assembly of micro parts based on three microscope vision systems is proposed to achieve complex micro parts assembly,The strategy uses the coarse adjustment process and the fine adjustment process to respectively detect and adjust the poses of the micro parts in the smallest and higher magnification,to realize the pose alignment and assembly for micro parts.An automatic zooming algorithm is proposed based on the image region of interest tracking technology,which can change the microscope vision system magnification automatically to ensure the feature area of the micro parts in the field of view.Through analysis of the image feature of the micro parts in the microscope vision system,the relative pose detection algorithm for micro parts is proposed based on image feature extraction.The relative pose detection of the micro parts in the three dimensional space was realized.The experimental results showed that the pose detection error was less than 0.5 degree,and the position detection error was less than 5 μm,which verify the effectivenees of the proposed algorithm.

关 键 词：位姿检测 自动变倍 装配策略 显微视觉 微装配 

分 类 号：TP211.4[自动化与计算机技术—检测技术与自动化装置]