Research On Embedded Face Detection System

Nowadays, along with the security of both society and production being paid more and more attention, the intelligent systems, which are based on video capture accompany with the elements of image processing, are under the spotlight. And the identification systems, which are based on the face recognition, have the best prospect in application. At present, most of the systems are based on PC, with some disadvantage, such as difficult to carry, high cost, big consumption of power, etc. While with the technology development of embedded system, the function and the arithmetic capability of the embedded systems are improved, this makes the embedded video capture and recognition system possible. Therefore, in this paper, the embedded video capture and recognition system are studied.In this paper, the characteristic of embedded system is studied, and the hardware environment of embedded video capture and recognition system is also built up. The embedded development board, which is based on ARM9 and the core chip is S3C2440, is utilized for the hardware platform of system. The module of video capturing is composed of USB camera and camera platform. The major functional part of camera platform is steering engine, which can control the rotation of camera in both vertical and horizontal directions. For the design of memory space, SD card is added on the basis of NAND Flash, and pictures are stored on SD card, aiming at enlarging the capacity of face database, meanwhile, providing easy access to the updating of face database.Embedded OS coupled with application program, which is popular recently, is utilized for the development mode of the software system in this paper. Meanwhile the definition of Embedded OS is elaborated, comparing the characteristics of different kinds of Embedded OS; embedded Linux operation system is selected. Embedded Linux operating system is successfully transplanted, through analyzing the component and transplanting method of embedded system.The application program includes two subroutines of camera flatform control and face detection. The driver of steering engine and the control program of camera flatform are achieved, providing the friendly control panel for user; the working process of face detection module is designed; video signal collection program is accomplished based on V4L; For the algorithm of face identification, in allusion to the characteristic of internal memory, a kind of algorithm of face detection with good performance and high operation efficiency is presented and achieved based on skin segmentation and AdaBoost.The system is tested and based on the result the future work is determined.