Synthetic Fingerprint Generation Using Generative Adversarial Networks: A Review

In this era of stringent privacy-related laws, the manual collection of fingerprints is a challenging task. This act as a deterrent for collecting large-scale database which is a prerequisite for implementing deep learning approaches in fingerprint-based applications. So, to overcome these challenges researchers came up with various synthetic fingerprint generation approaches using Generative Adversarial Networks (GAN). GANs are deep learning-based generative models which help in generating realistic-looking synthetic data. They help in the generation of fingerprints that are comparative in terms of quality, features and characteristics, with that of manually collected samples. The objective of this paper is to review the existing approaches based on GAN that are used for the synthetic generation of fingerprints. Critical investigation of underlying technological details of various GAN variants for generating synthetic datasets and their comparative analysis aims to assist researchers in the generation of apt data for designing advanced applications. Finally, an appraisal of various performance metrics is presented for the evaluation of the synthetic fingerprint quality to facilitate research.