Harnessing Deep Learning and GAN Technologies for Palmprint Recognition

Palmprints, along with face, fingerprints, and iris, are a common biometric used in security and forensic applications due to their precision, reliability, and cost-effectiveness. Palmprint recognition systems record characteristics like flexion creases, secondary creases, and ridges, with 30% of latent fingerprints identifying. This study explores palmprint identification technology, which has been developed and tested over 15 years on various image resolutions. The paper demonstrates the basic structure of palmprint and its types. It focuses on the challenges that are associated with the palmprint recognition system. The paper also discusses the palmprint recognition process that comprises of image acquisition, image preprocessing, feature extraction and the matching. The paper summarizes various palmprint databases and focuses on the features of deep learning models and GAN based models with their performance, used in palmprint recognition system. It compares various learning-based, deep learning and generative adversarial network methods. It aims to improve biometric recognition systems and guide future research by exploring the use of various recent generative models for synthetic biometrics.