Reducing False Positives in Minutia Detection by Using the Proposed Fingerprint Alignment Technique

Abstract

AbstractThe important step in automatic fingerprint recognition system is to automatically and reliably extract minutia from the input fingerprint images. In such recognition systems, the orientation of the fingerprint image has influence on fingerprint image enhancement phase, minutia detection phase and minutia matching phase of the system. The fingerprint image rotation, translation, and registration are the commonly used techniques, to minimize the error in all these stages of fingerprint recognition. In this work, we approached two methods by which the minutia is detected. In the first method, the minutias are detected without aligning the image. In the second method, the input image is aligned using the proposed k-means clustering based fingerprint image rotation algorithm and then the minutias are detected. This proposed rotation algorithm could be applied as a pre-processing step before minutia detection. In both the methods the images are enhanced using the proposed Gabor filter. Finally the results clearly show that the aligned images give more accurate true minutias then the unaligned images. Hence, the result will be better detection of minutia as well as better matching with improved performance.KeywordsFingerprint Image EnhancementRidge EndingsRidge BifurcationGabor Filterk-means clusteringFingerprint Image RotationAlignmentMinutia Detection