A Wrist-Worn Diffuse Optical Tomography Biometric System

We present a Diffuse Optical Tomography (DOT)based biometric system that uses interior anatomical information for better privacy and security instead of external traits such as face or fingerprint. The DOT system has a wearable form factor covering the lower forearm and the wrist, where anatomical structures in the optical path of the DOT optodes capture the unique internal patterns used for biometrics. Our DOT scanner is low-cost, using COTS near-infrared LEDs and sensors. Our design also incorporates wrist vein imaging as a secondary modality to supplement the DOT. This paper details the design of the DOT system and the ensuing machine-learning pipeline. We demonstrate the utility of the DOT as a stand-alone biometric modality and the efficacy of its fusion with wrist vein patterns. Our early experimental findings show promising results, using a pilot dataset to achieve an area under the receiver operating characteristic curve (ROC AU C) of 0.999138 and an equal error rate (EER) of 1.27% for the DOT modality. The AUC and EER were 0.999655 and 0.48% for the wrist vein imaging modality only and 0.99989 and 0.21 % for the fusion of both modalities.