Stereo Image Matching Based on SURF Descriptor
2015 ◽
Vol 1083
◽
pp. 142-147
Keyword(s):
In order to improve the robustness performance of SURF descriptor applied to stereo image matching, a new matching method is proposed. By using the ratio of minimum to second min Euclidean distance of corresponding features, we can get the coarse matching points aggregation. Then, the epipolar line is computed from calibration parameters. Correspondences are taken as correct correspondences, only if they fall into a small neighborhood of their epipolar line. Taken errors into account, the neighborhood is set (-3, 3). Using this restriction, we can get the correct matching points aggregation. The experimental results show that the correct matches and matching efficiency are better than RANSAC.
2013 ◽
Vol 39
(4)
◽
pp. 308-317
◽
2011 ◽
Vol 2011
◽
pp. 1-11
◽
Keyword(s):
Keyword(s):
2014 ◽
Vol 686
◽
pp. 348-353