A Fingerprint Segmentation Scheme Based on Adaptive Threshold Estimation

To achieve quick response in Automatic Fingerprint Identification System (AFIS), segmentation plays an important role in pre-processing step. The particular step filters out the Region of Interest (ROI) from a raw fingerprint image by segregating the unwanted portion that does not contain important image feature. Here we propose a pixel based scheme executes in three phases namely (a) enhancement of scanned image, (b) estimation of threshold to classify each pixel as a foreground or background pixel and (c) a post-processing step employing morphological filters to achieve an proper separation of the entire ROI. We implement and validate the proposed algorithm on various standard databases and the experimental results establish the efficiency of the method. We compare the proposed method with few of the existing approaches to establish the effectiveness and accuracy of the proposed algorithm.