Point pattern matching using TIN and affine invariants
Bitao Fu, Cheng Wang, Zhiyuan Zeng · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2007
How to perform initial match is a key problem of automatic image matching. This paper proposes a point pattern matching algorithm to match two point sets coarsely using affine invariants, area ratio of adjacent triangles. Its process is to constitute triangulated irregular network (TIN) by point sets, search the homonymous triangles by area ratio and affine transformation equations, and then establish initial mapping relations between two point sets, finally use least square matching to match them accurately. Experiments show that it can match two affine transformation point sets automatically without manpower-intervention and other auxiliary information.