Exploration On Face Recognition System Based On The Embedded Platform Of Image Acquisition

In recent years, people hope to find an efficient and reliable authentication schemes in the fields of video surveillance, intelligent entrance guard, human-computer interaction and so on. And the face recognition system is the first choice to solve the problem. But the portability and flexibility of traditional face recognition system is unsatisfactory, and the performance of face recognition system based on embedded platform is also not satisfactory due to its limited computational ability. So people need a more reasonable solution of face recognition system.Using both the powerful computing ability at the PC end and the flexibility of embedded devices, we propose a more reasonable solution of the face recognition system. We try to get images from the ARM embedded platform, and then these images are sent to PC end for face recognition through the network communication.This article explores the whole process of face recognition. In the embedded platform, we used Embest Edukit-Ⅲ experiment platform as the target machine. The development of target machine involves customization of BootLoader, modification and transplantation of the Linux kernel, construction of NFS system, invoking of camera driver and acquisition of image etc. And then we proposed the prototype of the related modules. In the PC end, we had studied the face recognition algorithms in depth and optimized feature extraction algorithm so as to fully extract effective face features, during which we introduced the process of theoretical derivation and verified the feasibility and effectiveness of the algorithm through ORL face database.