基于双目视觉测量系统的特征点匹配研究  

Research on feature point matching based on binocular vision measurement system

作　　者：赵航 孙冬梅 董清卿 吴潇灿 ZHAO Hang;SUN Dongmei;DONG Qingqing;WU Xiaocan(Intelligent Industrial Testing Technology Laboratory,College of Electrical Engineering and Control Science,Nanjing Tech University,Nanjing 211816,China)

机构地区：[1]南京工业大学电气工程与控制科学学院智能工业检测技术实验室

出　　处：《现代电子技术》2019年第23期154-157,162,共5页Modern Electronics Technique

基　　金：国家自然科学基金项目(51277092)：磁悬浮飞轮储能系统中陀螺效应的利用及运行控制研究~~

摘　　要：双目视觉测量作为一种新颖的测量方式,当前正在越来越多的测量中取代传统的接触式测量方法。但是由于双目视觉测量系统对测量精度和实时性要求较大,因此在分析椭圆拟合算法的基础上,通过筛选规则选取部分边缘点进行迭代最小二乘椭圆拟合以提取光斑中心坐标。依据系统对待测量笔9个圆形标志的位置反映的特点,设计一种基于RANSAC算法的特征点匹配方法。实验表明该算法可以满足双目视觉测量对系统的高精度和实时性要求。Binocular vision measurement technology as a novel measurement technology is gradually replacing the traditional contact measuring method. However,the binocular vision measurement system deserves high requirment on accuracy and real-time performance of the system. Therefore,a feature point matching method based on RANSAC algorithm was designed according to the position characteristics of 9 circular markers of measuring pen under test. On the basis of analysis of the elliptic fitting method,the center coordinates of facula are extracted by selecting some edge points by filtering rules for iterative least square ellipse fitting. The experiment results show that the method can meet the requirements of binocular vision measurement for high precision and real-time performance of system.

关 键 词：双目视觉测量 特征点匹配 位置约束 坐标提取 高精度测量 实验验证 

分 类 号：TN98-34[电子电信—信息与通信工程]