一种运动捕获系统中误匹配点的剔除算法  

Algorithm for eliminating mismatching points in motion capture system

作　　者：陈燕军[1] 郑江滨[1] 史文波[1] 李秀秀[1] 

机构地区：[1]西北工业大学计算机学院,西安710072

出　　处：《计算机应用研究》2009年第2期422-423,450,共3页Application Research of Computers

基　　金：国家"863"计划资助项目(2006AA01Z324);西北工业大学研究生创业种子基金资助项目(200849)

摘　　要：在特征点的提取阶段通过利用先验知识的迭代法得到二值化阈值,然后依据轮廓特征排除非特征点。对于特征点的匹配,首先根据极线约束得到初始候选点对,再通过松弛匹配算法和三角测量法剔出误匹配,得到最终的匹配点对;最后利用以上匹配算法给出了一种简便的多摄像机间特征点匹配的策略。实验结果表明算法匹配正确率高。Using the iterative method with prior knowledge,got the threshold at the stage of feature points,then excluded non-feature points based on contour feature.For matching feature points,obtained initial candidates points according epipolar constraint,then through the relaxation matching algorithm and triangulation measurement exclued mismatching points,it would to be last matching points.Finally,by the above matching algorithm described a kind of feature points matching strategy in a multi-camera system.Experi...

关 键 词：运动捕获 候选匹配点 松弛匹配 误匹配点 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]