A hybrid false minutiae removal algorithm with boundary elimination

We propose a hybrid false minutiae removal algorithm for accurate matching of low quality fingerprints. This algorithm is designed to be used in efficient security systems. It does not depend on filters like many present algorithms, rather it removes false minutia considering statistical information from a thinned binary image. Our proposed algorithm consists of two hybrid processes i.e. boundary elimination and false minutiae removal. It is found better than previous techniques and hence can be efficiently implemented in fingerprint identification systems. Our algorithm successfully removes prime false minutia like bridges, ridge breaks and spikes. The algorithm has been applied on FVC2002 and FVC2004. Experimental results, compared with other algorithms, show removal of almost 99% false minutia occurring in a dry, oily, dark or dirty image. Moreover, the algorithm hardly ever removes genuine minutia.