Fingerprint Recognition using Deep Learning

Abstract

Validity and consistency of fingerprint recognition has proven to be one of the most reliable methods for human identification. The fingerprint matching issue is conceived as an arrangement in which a model is created to learn to distinguish between a true and impostor pair of fingerprints. Previously, they used to exercise feature extraction prior to comparing a pair of fingerprints. Also, recently CNN has presented marvelous success for many images processing task. However, there are only a couple of attempts to develop a complete CNN method to influence challenges in the fingerprint recognition problem. We attempted to build a CNN-based fingerprint matching system in this research. The ability to learn fingerprint patterns directly from raw pixels in photos is a significant contribution of the technology. Incomplete and partial pairs of fingerprints were considered for feature extraction in order to achieve resilience and characterize commonalities broadly.