An improved method for singularity detection of fingerprint images

It is important for automatic fingerprint identification system to detect singularities of fingerprint images accurately and reliably. This paper proposes an improved method for singularity detection of fingerprint images. First, a fingerprint image is divided into non-overlapping blocks with the same size, the column which the core belongs to is acquired by detecting the horizontal part of the ridge, and then the row which the core belongs to is got. Also, the column and the row, which the delta belongs to, are acquired according to those acute variable ridge directions of left and right parts of the delta. Then, singularities are detected with Poincare index within the block image which the column and the row belong to. If no singularity is detected in the block image, search area is expanded to detect. At worst, the whole image is searched. Experimental results indicate that the improved method is effective and, even to low quality image, it still can detect the singularities reliably.