Design And Implementation Of Embedded Near-Infrared Face Recognition Platform

Face recognition technology is undoubtedly one of the most widely and the most mature technology in biometric identification technology. Traditional authentication methods, such as key, password, identity card, their performance about convenience and security can no longer meet the needs of social development. The face recognition technology is based on the difference between different people face, such as the location of eyes, the size of nose, the shape of mouth, by comparing the input face image or video stream with the information of the known face in database, to identify accurately the identification of the object.Currently, most of the face recognition system’implement is based on the PC platform.However, with the development of technology and our country pay attention to the security of information,the application about face recognition technology become more widely in some field,such as Criminal Investigation Identification, National anti-terrorism, Access security.Therefore,there has important significance to research and develop a face recognition system with fast speed and high accuracy.This article will introduce how to impelment an embedded face recognition system based on TI conpany’s OMAP3730processor.Firstly, near infrared face recognition system design, system functions to the hardware from the chip selection for top-level planning. For systems use two-way video case study using CPLD logic control dual camera image transmission and active near-infrared light exposure time, which not only makes the image signal processor OMAP3730has simultaneously handle two-way video capabilities, but also greatly reduces the active light source power. In addition, for OMAP3730dual-core processor, the paper introduces the ARM-based side UBOOT and LINUX kernel transplantation, focusing on analysis of the processor’s image signal processing module, and transplanted LINUX V4L2video driver. For DSP, TI’s DVSDK algorithm suite transplant NIR face recognition algorithm, and according to the actual effect is optimized. Finally, the paper analyzes the results of scientific and targeted approach proposed improvements, and achieved good results.Overall, this thesis design for embedded face recognition system face recognition algorithm, image acquisition and software interactions conducted in-depth analysis of the system and solve related problems, the successful realization of embedded near infrared face recognition platform, with a better performance.