一种基于角点引导的快速立体边缘匹配方法

被引:9
作者
李海超
张广军
机构
[1] 北京航空航天大学仪器科学与光电工程学院
关键词
立体视觉; 立体匹配; 快速边缘匹配; 角点约束; 边缘约束;
D O I
10.13700/j.bh.1001-5965.2007.05.013
中图分类号
TP391.41 [];
学科分类号
080203 ;
摘要
提出了一种新的基于角点引导的快速立体边缘匹配方法.在匹配过程中,为了保证其快速性引入了两种新约束:角点约束与边缘约束.首先,利用匹配的角点信息引导边缘的匹配,对如何利用已匹配的角点引导边缘匹配进行了详细的阐述,同时,在引导过程中提出了只利用8个方向的区域扩张方法,进一步保证快速性.其次,提出边缘约束实现点与点的匹配,使搜索范围仅限制在几个像素内,再利用极线约束实现点的匹配,必要时可结合灰度相关约束.结果表明:该算法取得比较好的匹配效果,引入的两种约束在很大程度上提高了速度和正确率,边缘点匹配的正确率高达97%以上.
引用
收藏
页码:557 / 560
页数:4
相关论文
共 8 条
[1]  
Computer vision:three-dimensional datafrom images. Klette R,Schluns K. . 1998
[2]  
Automatic matching and three-dimensional reconstruction of free-form linear features fromstereo images. Habib A,Lee Y R,Morgan M. Photogrammetric Engineering&RemoteSensing . 2003
[3]  
Effective parameters in search space reduc-tion used in a fast edge-based stereo matching. Moallem P,Faez K. Journal ofCircuits,Systems,and Computers . 2005
[4]  
A novel fast featurebased stereo matching algorithm with lowinvalid matching. Moallem P,Ashourian M,Mirzaeian B. WSEAS Transactions on Computers . 2006
[5]  
Contour matching using epipolar geometry. Han J H,Park J S. IEEE Transactions on Pattern Analysis and Machine Intel-ligence . 2000
[6]  
A computational approach to edge detection. Canny J. IEEE Transactions on Pattern Analysis and Machine Intelli-gence . 1986
[7]  
Automatic linematching and 3D reconstruction of buildings from multipleviews. Baillard C,Schmid C,Zisserman A,et al. IAPRS . 1999
[8]  
A robust technique formatching two uncalibrated images through the recovery of theunknown epipolar geometry. Zhang Z,Deriche R,Faugeras O,et al. Artificial Intelligence Jounr-nal . 1995