Fingerprint authentication on touch sensor using Phase-Only Correlation method

Fingerprint authentication becomes one of the solutions for security problems in smartphone. In the last few years, fingerprint-based authentication has been widely used and implemented to smartphone with a type of small area sensor called “touch sensor”. We present an approach that used phase component of image for matching called Phase-Only Correlation (POC) method that is highly robust against noise, brightness change, and image shifting. The proposed algorithm to solve partial fingerprint matching problem for small area matching against templates of a larger area problem consists of two steps; global search and local search. Global search aims to improve image acquisition, consists of several steps; rotation alignment, translational displacement alignment, and matching score calculation using BLPOC function with precise rotation. In local search step, more accurate match position from the result of the global search will be searched by shifting several pixels around the block. The authentication test using entire dataset of FVC2000 DB2 (set A) gives result of EER 1.7020%.