A pixel based segmentation scheme for fingerprint images

Abstract

Fingerprint segmentation, an important step in Automatic Fingerprint Identification System (AFIS) helps reduce the time of subsequent processing. It aims at separating the foreground region from the background. This paper presents a pixelwise segmentation scheme based on mathematical moment which provides a proper discrimination of the pixel intensities. A global threshold value is estimated from a set of local blocks with higher standard deviation. The relative local threshold values are derived subsequently to decide whether a pixel belongs to the foreground or background. Finally, morphological filtering is employed as post-processing step to identify the entire foreground region. The proposed method has been implemented and tested on a set offingerprint images and the experimental results visually establish the effectiveness of the method. Besides, a comparison with the existing methods is presented to verify the accuracy of the proposed algorithm.