A Hybrid Model for Fingerprint Recognition via LSTM and CNN

This paper proposes a Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) based hybrid model for enhancement of fingerprint recognition and matching from collective datasets. A CNN model is deployed for feature extraction learning from the source image data and Long Short-Term Memory for prediction in sequence at even time series which obtains optimized results against prevailing techniques with the base of specific metrics. The outputs of the hybrid approach demonstrate good recognition of fingerprint with accuracy of 98%. The proposed approach can potentially reduce the false acceptance and false rejection rates in fingerprint recognition systems, thereby enhancing their overall security and usability. The implication of using LSTM and CNN for fingerprint recognition is a significant advancement in biometric authentication technology with the potential for wide-ranging applications inclusive of law enforcement, border security, and access control systems.