Adaptive Linear Gabor Response Pattern for Palm-print Recognition

Abstract

The advancement of the information society has increased the demand for secure identity systems in a wide range of applications, including financial transactions, forensics, border control, computer security, and law enforcement. Biometrics is an alternative method for accurate and trustworthy authentication in a highly networked society. It is the science of using behavioural (gait, keystroke, signature, etc.) and physiological (face, ear, finger, palmprint, DNA etc.) attributes to validate an individual's identity. The special attributes on which the biometric recognition system is developed are known as modalities or identifiers. Recently, research in hand biometrics has shifted to examining the value of palm-prints in real-world contexts such as smartphones, ATMs, and notebooks. Palm prints are one of the most reliable biometrics available. The palm-print is considered as a bit of texture and texture-based feature extraction method has been presented during this work. This work proposes a palm-print recognition system that extracts textural characteristics by adjusting the optimum Gabor filter settings, named as adaptive linear Gabor response pattern (ALGRP). A distinct Gabor filter orientation is used to create a histogram for each sub-block of an image, and the sub-blocks of images are then concatenated to create a feature vector. The matching was made translation and rotation-invariant by using the normalized hamming distance. The proficiency of the proposed technique has been assessed on three publicly available palm-print databases. Test results exhibited that the method accomplished good performance with an EER of 0.094% and 0.13% on PolyU and IIT-Delhi databases respectively. We show that the proposed method outperforms the existing ones in terms of performance.