Personal verification using finger vein biometrics in FPGA-based System-on-Chip

As security and privacy protection have become important today, biometric-based personal verification has also increasingly gain popularity. However, biometric technology applies complex, compute-intensive image processing algorithms which demand to be executed on powerful computers for acceptable processing times. This prevents biometric systems to be used in applications that require low-cost embedded solutions. This paper presents a novel approach to personal verification that utilizes infrared finger vein biometrics implemented in an FPGA-based System-on-Chip platform. To improve performance, we apply the hardware-software co-design approach, using hardware acceleration for the image preprocessing tasks and image buffer management while leaving feature extraction and template matching to software. The system is prototyped on an Altera Nios2 FPGA development board running the Nios2-Linux RTOS at 100MHz clock. The proposed finger vein biometric system achieves a matching accuracy of EER (average) of 0.08% and a verification processing time of 1.27 seconds, demonstrating that the design approach taken can be effective for embedded vein biometric authentication system.