COMPARISON OF ERROR METRICS IN MATCHING ALGORITHMS OF IMAGES BY SURF DETECTOR
Keywords:error metric, feature point, homography matrix, normalized cross-correlation, Speed-Up Robust Feature
AbstractSpeed-Up Robust Feature (SURF) method is used to detect feature points of images. The analyses ofmatching algorithms of feature points is done. The comparison of different error metrics by their accuracy andcomputing efficiency is provided on the series of test images for basic transforms like scaling, rotation and shifting.
H. Bay, T. Tuytelaars, L. Van Gool. SURF: Speeded up robust features. European Conference on Computer Vision, Vol. 1, pp.404–417, 2006.
Y. X. Chen, Q. S. Sun, H. Y. Xu, et al. "Matching method based on SURF algorithm and RANSAC algorithm" // Journal of Frontiers of Computer Science and Technology, vol. 6, 2012. - p. 1-9.
D. G. Lowe “Distinctive Image Features from Scale-Invariant Keypoints” // International Journal of Computer Vision, vol. 60, no. 2, pp. 91–110, 2004.
M. Marius, G. L. David "Fast approximate nearest neighbors with automatic algorithm conjugation" // Proc. Computer Vision Theory and Applications' 09, 2009, pp. 331-340.
K. Mikolajczyk, C. Schmid “A Performance Evaluation of Local Descriptors” // Proceedings of Computer Vision and Pattern Recognition, June, 2003.
J. Shi, C. Tomasi “Good features to track” // IEEE Conference on Computer Vision and Pattern Recognition (CVPR94), 1994.