Minutiae Triple Correlation: A Translation Invariant Fingerprint Representation

In this paper, we introduce a novel translation-invariant minutiae representation based on triple correlation for fingerprint biometric recognition. In contrast with other fingerprint representations with translation-invariant characteristics, the proposed one does not lose any information about the original trait, with the only exception of its absolute position. Nevertheless, triple correlation would be of little use, in its native form, for signal processing applications, due to its high-demanding memory requirements. By exploiting the intrinsic sparsity of the treated minutiae, we propose a sparse implementation of the employed triple correlation, allowing its use in practical application and the definition of a suitable comparison algorithm here introduced. The performed experimental tests show the suitability of the proposed triple-correlation representation for fingerprint minutiae comparison.