SUBIMAGE REGISTRATION BASED ON CANDIDATE SELECTION AND MULTIPLE CORRESPONDENCES

Identifier:  TFM:27

Handle: https://hdl.handle.net/20.500.11797/TFM27

Authors:  Amado Cepeda, Victoria

Abstract:
SUBIMAGE REGISTRATION BASED ON CANDIDATE SELECTION AND MULTIPLE CORRESPONDENCES In some fields like forensic research, experts demand that a found sample of an individual can be matched with its full counterpart contained on a database. This found sample might present several characteristics that make this matching more difficult to perform, such as distortion and most importantly, a very small size. Several solutions have been presented intending to solve this problem, however, big computational effort or low recognition rate is obtained. In this master thesis, we present a fast, simple and efficient method to relate a small sample of a partial palmprint to a full one using elemental optimization processes and a voting mechanic. Experimentation shows that our method performs a higher recognition rate than the state of the art Hough method, when trying to identify palmprint samples with a radius as small as 2.64 cm.