基于RANSAC算法的基本矩阵估计的匹配方法.pdfVIP

167 122730( 2006 0420066204 RAN SAC 1 2 1 , , ( 1. 上海师范大学数理信息学院, 上海, 200234; 2. 上海电机学院 电子信息学院, 上海, 200240 , , Samp son RAN SAC( R an2 dom Sampling Con sen su s , ; ; RAN SAC ; Sampson T P 3 17. 4 A The Matching Method Base d on RANSAC Algorithm for Estimation of the Fundamental Matrix 1 2 1 Shan Xin , W ang Yaoming , Dong Jianping ( 1. Mathemat ics and Sciences College, Shanghai Norm al University, Shanghai, 200234; 2. School of Electronic Inform at ion , Shanghai Dianj i U niversity , Shanghai, 200240 Ab stract The fundament al matrix contain s all of a cam er a. s inner and out er paramet ers of a cam2 era. So the est imat ion of the fundament al m atrix is a crucial problem of computer vision . In this pa2 p er, we introduce the theorem of the epipolar geom etry and the fundam ent al m atrix . T o solve the problem of false mat ches, we also introduces RAN SAC algorithm based on Samp son dist ance. Experi2 ments on featur e det ect ion and m atching as well as the est imation of the fundament al m atrix show that the pr oposed method is effect ive. Key word s fundament al matr ix ; epipolar geom etry ; r andom sampling consensus( RAN SAC al2 gorithm; Samp son dist ance , , , , : 20062 062 01 : ( 0 1D022 1 : ( 19812 , , , : , [ 1] 2 Iu ( x) I uv ( x) : C ( x ) = 2 , Iu ( x) I v ( x ) H arr is I uv ( x ) I v ( x) , I uv ( x)