Fingerprint segmentation based on ridge search

Fingerprint image segmentation is an important preprocessing step in Automatic Fingerprint Identification System (AFIS). A well designed fingerprint segmentation method can improve the accuracy of the feature extraction. The traditional gray variance segmentation method is simple and widely used, but it can hardly segment fingerprints with low contrast or high noise. A novel segmentation method based on ridge search is presented in this paper. The new ridge coherence and the ridge distance variance of fingerprints are obtained through ridge search for segmentation. A hybrid method is also proposed by combining the merits of the traditional gray variance and our novel segmentation method. Experimental results on typical fingerprint images indicate that the proposed method has advantages in poor quality fingerprints especially when the texture can not be well identified. The equal error rate (EER) is used to evaluate the performance of these three segmentation methods on public databases FVC2000 DB3, FVC2004 DB2 and DB4. EER is reduced when using the novel segmentation and the hybrid method relative to the traditional gray variance method.