Abstract

Recently, biometric recognition has emerged quite popular in various advanced human authentication systems like secure access
control systems, forensic applications and criminal identification systems etc. It is a technique to identify a human automatically
with the help of a computational algorithm using the biometric features present in database. Finger vein features based recognition is
also a type of the biometric authentication systems used in this study. In the earlier work, Corner detection algorithm is utilized for
the extraction of features, so called corner points from finger vein image, and pattern matching was carried out according to
differences between corners denoted as points employing the neural network classifier. But, if finger-vein images quality are not
up-to-the mark, segmentation errors might crop up while the vein patterns are extracted (binarized). In order to resolve this problem,
texture feature extraction is performed with the help of grey level co-occurrence matrix in this work. For improving finger-vein
images perceptibility during the pre-processing of the feature extraction technique, scattering removal and vein enhancement
techniques have been also presented in this study. This research work uses thinning and denoising for the removal of the
unnecessary pixels in the vein pattern. Corner detection is carried out with the help of the Harris corner detection operator and the
tracking of the finger vein branches will be carried using Improved Fuzzy c-means Clustering algorithm. At last, Morphological
dilation and dot product operations are carried out on extracted finger vein to increase pixel quality and to get the real or fake finger
vein image depending on User-defined threshold.