Fingerprint images segmentation based on fuzzy C- mean theory and statistical features

Fingerprint segmentation is a crucial and important step of image processing in automatic fingerprint identification. Because, it is very important for alright fingerprint features extraction, such as, singular points, bifurcation and ridge ending minutia’s. The aim of the segmentation of fingerprint is to extract the interest area (foreground) and to exclude the background regions, in order to reduce the time of subsequent processing and to avoid detecting false features. This paper presents a new approach of fingerprints segmentation. This approach is based on variance image and combined fuzzy C-mean algorithm with the statistical features. Fingerprint segmentation results from the proposed method are validated and the accuracy of segmentation sensitivity for the test data available is evaluated. We have tested this technique on more than 1000 images fingerprint taken from “CASIA Fingerprint Image Database Version 5.0” (CASIA-FingerprintV5). Then a comparative study with the existing techniques is presented. The experimental results demonstrate the superiority, the effectiveness and the robustness of the proposed method.