机器视觉技术中一种基于反对称矩阵及RANSAC算法的摄像机自标定方法  

An Approach of Camera Self-Calibration Based on Skew-symmetric Matrix and RANSAC alogrithm

作　　者：王赟[1] 

机构地区：[1]新乡学院机电工程学院,新乡453003

出　　处：《现代制造技术与装备》2015年第4期92-94,共3页Modern Manufacturing Technology and Equipment

摘　　要：介绍了一种摄像机自标定方法,该方法通过匹配的特征点建立标准矩阵后,利用反对称矩阵的性质,将标准矩阵表达式分解成6个约束方程,通过其约束关系得到摄像机内外参数。同时采用了RANSAC算法从检测到的特征点中排除奇异的特征点,对数据集进行筛选,以此提高匹配点的准确度和标定的精度。实验表明该方法能根据真实视频获得摄像机内外参数,能够较好的应用于机器视觉领域。This paper describes a self-calibration method . After establishing fundamental matrix by using matched feature points , six constraints equations was founded from the fundamental matrix based on the character of the skew-symmetric matrix . . Then the intrinsic and ex-trinsic parameters can be determined through the relation of the set of constraints . Ransac method was adopted to exclude the singular points from detected feature points , therefore improve the accuracy of feature matching and camera calibration . Experimental results for real video showed that this method can effectively acquire the intrinsic and extrin-sic parameters , and it can be applied into computer vision field .

关 键 词：摄像机自标定 基本矩阵 反对称矩阵 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]