视觉测量中椭圆自动检测与定位方法  

Automated ellipse detection and location method in 3D visual inspection

作　　者：韩建栋[1] 杨红菊[1] 吕乃光[2] 

机构地区：[1]山西大学计算机与信息技术学院,计算智能与中文信息处理教育部重点实验室,太原030006 [2]北京信息科技大学光电信息与通信工程学院,北京100192

出　　处：《计算机工程与应用》2011年第17期169-171,191,共4页Computer Engineering and Applications

基　　金：国家自然科学基金No.50675015;北京市属市管高等学校人才强教计划资助项目(No.PXM2007_014215_044674);北京市重点实验室(机电系统测控)开放课题资助项目(No.82073011)~~

摘　　要：针对视觉测量中椭圆目标的检测效率与定位精度较低的问题,提出一种高效的椭圆识别与定位方法。该方法利用回光反射控制点的强反射特性,提取目标的边缘信息,对连续的边缘进行分组,利用椭圆长轴信息对每组边缘数据进行椭圆识别,采用基于变量含误差模型的椭圆拟合方法,实现椭圆目标中心的精确定位。实验证明,该方法具有快速、自动化、定位精度高、鲁棒性好等优点,在视觉测量中具有广泛的应用前景。In order to solve the problems in 3D vision measurement such as low inspection efficiency and low location precision and so on,an efficient method of ellipse detection and location is proposed.Using high-reflective characteristic of the retro-reflective targets,the Canny detector is applied to the gray level image in order to obtain a binary edge-map.Edge contours points are extracted from the edge-map and divided into several groups according to the continuity.Ellipse objects can be recognized by using major axis information of an ellipse.An ellipse fitting method based on the Heteroscedastic Errors-In-Variable（HEIV） model is applied to realize the precision locating of the ellipse objects.Experimental results show that the method has such advantages as fast speed,automation,high precision and good robustness and can be extensively applied to the vision measurement.

关 键 词：视觉测量 椭圆识别 边缘提取 椭圆拟合 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]