Performance Evaluation of Biometric Authentication and Classification Using Deep Learning Approach

Biometric authentication and categorization of fingerprints and retina images have been a major difficulty in intelligent computing in several industries over the last decade. The main motivation of the current research work is the increasing number of people who are attacked by hackers every day especially in the field of the banking industry. So there is a need to develop intelligent computing techniques for accessing and identifying the person more accurately for secured transactions. This research work aims to classify the fingerprint and retina image and measure the performance metrics using these intelligent computing techniques. The Biometric Authentication and Classification Using Deep Learning Approach contribution was based on Fragment Jaya Whale Optimizer with Deep Convolutional Neural Network (FJWO-DCNN) for achieving a good recognition rate using the optimal theoretical features of Biometric images. The generated theoretical features ensure the effectiveness of the optimal solution that yields to detect the Fingerprintand retina image more accurately. Then the extracted theoretical features and the original image were given as the input to different level classification strategy which were performed using Deep Convolution Neural Network (DCNN) classifier, which was trained by the proposed Fragment Jaya Whale Optimization (FJWO) algorithm. The Casia fingerprint and the STARE database are used in this investigation. Finally, employing this hybrid approach, the results were obtained with a higher performance value. This proposed work achieves a higher performance percentage in terms of accuracy.