Efficient cancellable multi-biometric recognition system based on deep learning and bio-hashing

Cancellable biometrics have been enrolled in several applications such as cloud computing and cyber security. This makes researchers investigate their approaches in this field. This paper presents a Cancellable Multi-Biometric System (CMBS) based on deep image style transfer and a fusion process. The main contribution is cascading style transfer processes of the human biometrics including fingerprint, finger vein and face images. Then, a fusion process is carried out on the style transferred images. The generated cancellable templates are evaluated by both visual and statistical analysis. The results of the proposed system show superior performance in terms of Area Under the Curve (AUC) and encryption quality assessment with Structural Similarity Index Measure (SSIM), Number of Changing Pixel Rate (NPCR) and other quality indices. Furthermore, the generated templates are digested using hashing algorithms including SHA-224 and SHA-256. The proposed system is compared to the works in the literature. The comparison reveals that the proposed system has a superior performance compared to other previous ones. Hence, it can be used in biometric authentication in cloud systems.