基本矩阵的5点与4点算法_英文_吴福朝.pdfVIP

29 　2 　　　　 Vol.29, N o.2 200 3 3 ACT A A U TO MA T ICA SIN ICA M ar., 2003 5-point and 4-point Algorithm to Determine of the Fundamental Matrix1) WU Fu-Chao 　　HU Z han-Yi (National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, Beijing　100080) (E-mail:{cw u, huzy}@) Abstract　The undamental matrix encapsulates all the in ormation between tw o images, and plays a very important role in camera calibration and 3D reconstruction.In this paper, the ollowing conclusions have been rigorously proved:I the camera motion is o a pure translation, then g iven 5 point correspon- dences across tw o images, the undamental matrix can be linearly determined i our correspondences o the 5 ones are rom coplanar space points (called coplanar co rrespondences).In addition, we show that i the distortion actor in the pinhole camera model is null, then the undamental matrix can be linearly determined by only these 4 coplanar cor respondences .T o our knowledge, such results have not been re- ported yet in the literature . Key words　Fundamental matrix, homography , camera intrinsic parameters 1　Introduction T he epipolar geometry is the undament al const raint between tw o images.It is indepen- dent o scene structure, and depends only on the camera internal parameters and their relative pose .T he undamental mat rix encap sulates this epipolar geomet ry, and plays a very impo rtant role in 3D computer vision[1, 2, 3] , such as camera calibration , 3D reconstruction.The