Reasearch And Application On Embedded Face Recognition System

With the development of computer technology, the image information processing andrelated applications are becoming more and more common in today’s social life. The ARMEmbedded system as a kind of miniaturization, low power consumption and easy to operateparticular computer system, gradually move into all aspects of social production, and itsapplication is increasingly wide. On this basis, the use of the ARM Embedded System forimage information processing, greatly enhances the flexible, interactive, data processingcapabilities of image processing system. The application of the image information processingtechnology is more extensive and popular. Biometric field as an embedded image processingan application of the direction, and gradually become a hot topic in recent years, attracted theattention of the domestic and international numerous scholars and engineers.This thesis researched a image information processing system based on the ARMembedded system and realized face detection on it which combined with the relevantknowledge of the current popular face recognition. Established embedded image developmentplatform and laboratory face database, which laid a solid foundation for the follow-up study.Firstly, from the perspective of embedded, this paper explained the design process of thehardware and software system platform, including the development of embedded system andthe selection of embedded hardware based on the requirements of the embedded systems.Complete port of the kernel of Linux operating system, BootLoader, file system and the twoauxiliary library function OpenCV and Qt in the hardware platform Tiny210. It provides apowerful tool for the development image information processing system development.Secondly, focus on USB camera video capture on the embedded system developmentprocess, its main processes include four major parts which are drive configuration of theLinux kernel, driver Video4Linux2application, how to call Qt to display image on the screenand PC display. Then, complete software development about embedded ARM systems USBcamera video image acquisition and display, and establish a development platform forembedded system image processing.From the point of view of texture feature, a face recognition algorithm based on thecurvelet domain kernel principal component analysis is proposed. Study on algorithm of facedetection based on the haar characteristics. Design a face recognition system, which includesdetection module, acquisition module, identification module, and display module. In thesoftware program, gather the Laboratory face database, and realize the face detection andrecognition system successfully.Finally, this thesis summarizes the research achievements and deficiency of this system,and puts forward the future direction of the research.