An optimal minutiae-based combining method for multiple fingerprint templates

Abstract

This paper proposes a novel method for combining multiple fingerprint templates to improve the successful matching rate in fingerprint verification systems. Firstly, SVD is used to optimal aligning of two minutia sets. Then, based on the newly obtained overlapping area, we introduce a post-processing step to collect previously missing minutiae pairs due to nonlinear distortion. The overall advantages of our method are compared and tested in standard FVC databases in terms of RMS error, the average number of aligned minutia pairs and ROC curve of verification systems.