Clustering Based Fingerprint Indexing Using Triangle Spiral

The voluminous biometric databases make it computationally difficult to match a query against each enrolled image in the database as is required in identification. To make the process efficient, various indexing techniques have been developed. We propose a novel fingerprint indexing scheme using dynamic clustering and m-ary trees in this paper. We construct a triangle spiral for each fingerprint image from which we obtain O(n) triangles unlike many other indexing approaches. These triangles are used to extract robust translation and rotation invariant features. Experiments with the benchmark databases confirm the superiority of our approach to the other existing techniques.