Template synthesis and image mosaicking for fingerprint registration: an experimental study

Fingerprint registration becomes an important issue for the success of reliable fingerprint verification using small solid state fingerprint sensors. Current major registration techniques include template synthesis and fingerprint image mosaicking. Template synthesis merges the fingerprint features while image mosaicking merges the fingerprint images to generate a composite fingerprint image from features of the images. In this paper, these two techniques were studied and compared under a unique framework to review their characteristics operating under fingerprint images of different sizes. Experiments show that both techniques could have their own merits. In particular, we also present a fingerprint oriented ICP (iterative closest point) algorithm that aligns the images based on their extracted minutiae.