子像素匹配策略的对极几何估计算法The estimation algorithm of epipolar geometry based on sub-pixel match strategy
山海涛,马淑宇,郝向阳,李海滨
摘要(Abstract)：
基本矩阵在计算机视觉中扮演着十分重要的角色,对极几何(即核线几何)的估计过程就是求解基本矩阵的过程,所以如何提高对极几何的估计精度一直是计算机视觉中的一个研究热点。本文通过提取子像素特征点,并经高精度影像匹配,最终用子像素匹配点来实现对极几何的稳健估计。实验表明,该计算法在平均余差和平均对级距离上都明显小于基于整像素级的对极几何估计结果,有效改善了对级几何的估计精度和稳定性。
关键词(KeyWords)： 计算机视觉;基本矩阵;对极几何;子像素匹配
基金项目(Foundation):
作者(Author): 山海涛,马淑宇,郝向阳,李海滨
DOI: 10.16251/j.cnki.1009-2307.2010.06.072
参考文献(References)：
- [1]Faugeras O.Three-Dimensional Computer Vision:A Ge-ometric Viewpoint[M].England London Cambridge,Massachusetts:The MIT Press,1993.
- [2]Deriche R.and Zhang Z.and Luong Q.T.and Faugeras O.Robust recovery of the epipolar geometry for an uncal-ibrated stereo rig[C]//Proceedings of the3rd Europe-an Conference on Computer vision(ECCV'94).Germa-ny Berlin:Heidelberg Springer-verlag,1994:567-576.
- [3]Faugeras O.What can be seen in three dimensions with an unclibrated stereo rig[C]//Proceedings of the 2nd European Conference on Computer vision(ECCV'92).Germany Berlin:Heidelberg Springer-verlag,1992:563-578.
- [4]Quan Long and Kanade Takeo.Affine structure from line correspondences with uncalibrated affine cameras[J].IEEE Trans.on Pattern Analysis and Machine Intelli-gence.1997,19(8).
- [5]Hartley R,Zisserman.A Multiple View Geometry in Computer Vision[M].Cambridge:Cambridge Univer-sity Press,2000.
- [6]Longuet-Higgins H C.A computer algorithm for construc-ting a scene from two projections[J].Nature,1981,293.
- [7]Hartley R.In defence of the8-point algorithm[C]//Proceedings of the5th ICCV.USA Cambridge,1995:882-887.
- [8]Wang Wei,Shen Peiyi,Wu Chengke.The redundant problem of the collinear points in the fundamental matrix calculation[J].Journal of Xidian University,1998,25(4).
- [9]Zhengyou Zhang,Rachid Deriche et al.A robust tech-nique for matching two uncalibrated images through the recovery of the unknown epipolar geometry[J].Artifi-cial Intelligence1995,78:87-119.