Deep Learning-Based Hyperspectral Multimodal Biometric Authentication System Using Palmprint and Dorsal Hand Vein

Abstract

Biometrics recognition has been widely adopted for authentication, i.e., face, iris, palmprint, palm vein, dorsal hand vein, and fingerprint. Particularly, palmprint has stable and rich characteristics, such as textures, wrinkles, and local direction features. Similar to palmprint, dorsal hand vein also has stable characteristics that do not change with age, which can effectively protect against spoofing attacks and impersonation. Recently, hyperspectral imaging technology has been utilised in biometric authentication research. If the hyperspectral technology is utilised in both palmprint and dorsal hand vein imaging, more discriminative and living characteristics can be acquired to enhance the security. In this work, we designed a novel near-infrared multimodal hyperspectral palmprint and dorsal hand vein authentication system. Based on the designed system, we constructed a hyperspectral palmprint and dorsal hand vein dataset to verify the recognition accuracy. The effectiveness of the proposed system has been proven as a real-world biometric security authentication application.