PERFORMANCE ANALYSIS OF A SECURE FINGER VEIN RECOGNITION SYSTEM USING HYBRID FEATURE EXTRACTION AND FEATURE SELECTION

Abstract

Most common physical biometric for authentication purposes are the fingerprint, hand, iris, face, discern vein, DNA and voice. The advantage claimed with the aid of biometric structuresis they can set up an unbreakable one-on-one correspondence amongcharacter and a bit of data. Biometrics provides
authentication advantagesacross the spectrum, from IT companies to end users, and from UAID gadgetdevelopers to UAID gadget users. A good biometric is characterized with the aid of use of a characteristicit highly unique and precise:sothe possibility of any two human beings having an equivalent characteristics are going to be minimal, stable: so that the characteristicwould notchange over time, and be effortlessly acquired: as a way todelivercomfort to the user, and prevent misrepresentation of the features. Fingerprint recognition is that the oldest technique of biometric authentication. In those instances the fingerprint identitytechniquebecame used, with the name as Actyloscopy. In this work, Hybrid Feature Extraction (HFE) with security based biometric
system will be introduced for evaluating the performances. HFE contains Histogram of Oriented Gradients (HOG), Stationary Wavelet Transform (SWT), Grey Level Co-occurrence Matrix (GLCM), Local Binary Pattern (LBP), and Principle Component Analysis (PCA). For Feature selection, KNN based
Genetic Algorithm (GA) is used and the classifier used in this proposed methodology is error correcting code based SVM (ECOC-SVM). Finally, the performance parameters are calculated in terms of such as accuracy, precision, recall, sensitivity, specificity, false acceptance rate and false rejection rate.