Cancelable biometric Iris Authentication based on Wavelet transforms and Double Random Phase Encoding for Cloud Services

Abstract

Cancelable biometric authentication represents essential progress in protective biometric methods, which protect private data but let users revoke templates. The proposed work combines wavelet transforms with Double Random Phase Encoding (WT-DRPE) along with Local Binary Patterns (LBP) for iris recognition to improve feature extraction methods while maintaining cryptographic non-invertibility techniques. The wavelet transform analyzes the iris image by separating it into frequency bands before the high-frequency parts receive DRPE-based encoding by applying two random phase masks. LBP retrieves iris texture information that converts into a secure, cancelable feature vector by applying random projection and bio-hashing. The research compares the proposed method against GWT-AES and RSA techniques in combination with chaotic encryption through analysis of Authentication Time and Extracted Feature Time alongside Genuine acceptance rate (GAR), False acceptance rate (FAR), and Equal error rate (EER). The proposed method reaches its efficiency mark by testing it on two datasets from the CASIA Iris V4 and IITD iris databases. Testing confirmed that the proposed system delivered outstanding accuracy with minimum EER levels, optimized computational speed, and surpassed existing techniques for unlinkable and revocable template operations.