A novel fingerprint indexing scheme using dynamic clustering

Abstract

With the increase in adoption of biometric data for various identification processes, the need for efficient indexing techniques is also on the rise as it is computationally impossible to match the query template against every enrolled entry in the database. In this paper, we have proposed two robust feature extraction techniques from fingerprint images followed by an indexing method based on clustering and m-ary trees. We extract robust translation and rotation invariant features from O(n) triangles obtained from the triangle spiral constructed for every fingerprint. The other method involves construction of M rectangles around every reference minutiae and calculation of features based on neighboring relation. Experiments with the benchmark databases confirm the superiority of our approach to the other existing techniques.