A DoG based Approach for Fingerprint Image Enhancement

Abstract

Fingerprints have been the most accepted tool for personal identification since many decades. It is also an invaluable tool for law enforcement and forensics for over a century, motivating the research in Automated fingerprint-based identification, an application of biometric system. The matching or identification accuracy using fingerprints has been shown to be very high. The theory on the uniqueness of fingerprint minutiae leads to the steps in studying the statistics of extracting the minutiae features reliably.
Fingerprint images obtained through various sources are rarely of perfect quality. They may be degraded or noisy due to variations in skin or poor scanning technique or due to poor impression condition. Hence enhancement techniques are applied on fingerprint images prior to the minutiae point extraction to get sure of less spurious and more accurate minutiae points from the reliable minutiae location.
This thesis focuses on fingerprint image enhancement techniques through histogram equalization applied locally on the degraded image. The proposed work is based on the Laplacian pyramid framework that decomposes the input image into a number of band-pass images to improve the local contrast, as well as the local edge information. The resultant image is passed through the regular methodologies of fingerprint, like ridge orientation, ridge frequency calculation, filtering, binarization and finally the morphological operation thinning. Experiments using different texture of images are conducted to enhance the images and to show a comparative result in terms of number of minutiae extracted from them along with the spurious and actual number existing in each enhanced image. Experimental results out performs well to overcome the counterpart of enhancement technique.