ENHANCED FACE RECOGNITION FOR SECURE DRIVER’S LICENSE DATA RETRIEVAL USING MULTI-LEVEL RED DEER CASCADE CONVOLUTIONAL NEURAL NETWORK

Abstract

Extracting driver's license information using facial recognition presents significant challenges, including handling diverse lighting conditions, face orientations, and occlusions while ensuring data security. Traditional identity verification methods that rely on physical documents or manual checks are prone to inaccuracies and security vulnerabilities. This study addresses the challenge by proposing a robust, automated system for face detection and recognition, aimed at secure retrieval of driver’s license data. To tackle the issues of varying illumination and facial appearances, they apply a pre-processing step that utilizes the Integral Normalized Gradient Image (INGI) method, transforming images into an illumination-insensitive format. Our system leverages a Multi-level Cascade Convolutional Neural Network (MLC-CNN) for face recognition. The CNN parameters are optimized using an enhanced Red Deer Optimization (RDO) algorithm to ensure high accuracy. The driver’s license details are then retrieved from the database upon successful facial recognition. Our approach demonstrates superior performance compared to existing methods, achieving enhanced accuracy and reliability under challenging conditions. This work provides a secure, efficient solution for facial recognition-based data retrieval, contributing to safer identity verification systems for driver's licenses. The proposed method achieved outstanding performance with an accuracy of 98.9%, precision of 99.4%, recall of 98.4%, and an F-measure of 98.9%, surpassing other approaches in terms of overall effectiveness.