Palmprint Based Human Recognition in Unconstrained Environment

Palmprint has emerged as one of the futuristic biometric traits for human recognition. It is texture rich and provides a large area for image acquisition. It can provide higher accuracy even with low-quality images. However, serious challenges are posed when the palmprint acquisition is in an unconstrained environment where a user can present his palm in many different ways. This makes segmentation of the region of interest (ROI) difficult, leading to lower recognition accuracy. This paper presents a palmprint based human recognition system by designing an automatic unconstrained palmprint acquisition device using low-cost scanners. Performance of the system is evaluated on an in-house palmprint database that has been collected from 100 users using the proposed prototype. The system has achieved recognition accuracy of 93.3% for rank3 comparison. The Correct Recognition Rate (CRR) and Equal Error Rate (EER) of the system have been found to be 92.3% and 6.8% respectively.