An effective algorithm for low quality fingerprint segmentation

In this paper, we have proposed a novel algorithm for fingerprint segmentation. Firstly, we originally use the method of gradient projection to eliminate those background regions which have smaller gray contrast, and then obtain the approximate foreground region of the fingerprint image. In addition, we adopt the gradient coherence criteria to exclude the regions which contain irregular stains and smudges. Finally, we creatively carry out morphological operations on the edge image of fingerprint to get the exact foreground region of the fingerprint, through which we could rule out blur parts in the fingerprint image and get a fingerprint foreground which has smooth contour. Experimental results show that our algorithm is effective and robust for most fingerprint images even those of low quality.