Wide baseline matching through homographic transformation

被引：2

作者：

Elias, R ^{[1
]}

机构：

[1] Univ Ottawa, SITE, VIVA Res Lab, Ottawa, ON K1N 6N5, Canada

来源：

PROCEEDINGS OF THE 17TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION, VOL 4 | 2004年

关键词：

D O I：

10.1109/ICPR.2004.1333721

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

This paper discusses the wide baseline matching problem where the camera parameters are known up to an error factor and the ground surface is considered planar Junctions of different shapes and orientations are detected. Homographic correlation, an invariant measure to projective transformation, is utilized using reconstructed planes made of detected junctions. Two variants of homographic correlation are proposed showing that this approach outperforms non-homographic correlation.

引用

页码：130 / 133

页数：4

共 8 条

[1] Sequential updating of projective and affine structure from motion
Beardsley, PA
Zisserman, A
Murray, DW
[J]. INTERNATIONAL JOURNAL OF COMPUTER VISION, 1997, 23 (03) : 235 - 259
[2] Falkenhagen L., 1994, IMAGE PROCESSING BRO, P115
[3] FAUGERAS O, 1996, 3 DIMENTIONAL COMPUT
[4] Hartley R., 2000, MULTIPLE VIEW GEOMET
[5] LAGANIERE R, 1996, CRPRRT9603 EC POL MO
[6] LAGANIERE R, 2004, P IEEE INT C AC SPEE
[7] Wide baseline stereo matching
Pritchett, P
Zisserman, A
[J]. SIXTH INTERNATIONAL CONFERENCE ON COMPUTER VISION, 1998, : 754 - 760
[8] TUYTELAARS T, 2001, UNPUB INT J COMP JUL

← 1 →