A novel compatible fingerprint algorithm based on multi-region alignment

Most minutiae-based matching algorithms consist of two phases, local match and global match. In the local phase, some corresponding pairs are obtained by comparing the affine-invariable features of minutiae. And then two images are aligned based on the candidate pairs. However, some spurious candidate pairs and the large nonlinear deformation in images lead to the failure in global match. In this paper, we proposed a novel minutiae-based matching scheme which insert a filtering step after the local match to discard the incompatible pairs and renovate the global match by dividing the whole image into small areas according to the location of the candidate pairs. Results on databases of FVC2004 validate our algorithm.