“Is there a transformed (rotated, translated, and scaled) subset of some model point-set which matches a subset of the scene point-set.” [4]
Reference frame -> a 2D hash table Local affine frame [2] uses triplet of non-collinear points. The 3 points are used to compute an affine transformation between a canonical frame (0,0,1), (0,1,1), (1,0,1).
min-hash and geometric min-hash
Geometric min-hash [3]
Feature map hashing [4]
[1] Wolfson Rigoutsos. "Geometric Hashing: An Overview." Computational Science & Engineering, 1997. [2] Chum Matas, "Geometric Hashing with Local Affine Frames." CVPR 2006 [3] Chum Perdoch Matas, "Geometric min-Hashing: Finding a (Thick) Needle in a Haystack." CVPR 2009. [4] Avrithis Tolias Kalantidis, "Feature map hashing: sub-linear indexing of apperance and global geometry." ACM Multimedia, 2010. [4] Y. Lamdan and H. Wolfson, “Geometric hashing: A general and efficient model-based recognition scheme,” ICCV 1988.
https://github.com/RSIA-LIESMARS-WHU/LSHBOX https://code.google.com/p/caltech-image-search/
