基于DCT的人脸识别算法在嵌入式系统的并行实现  

Parallel Implementation of Face Recognition Algorithm on Embedded System

作　　者：汪伟[1] 杨新[1] 

机构地区：[1]上海交通大学图像处理与模式识别研究所,上海200030

出　　处：《计算机应用研究》2005年第6期226-228,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(30170264);国家基础研究计划资助项目(2003CB716104)

摘　　要：应用双DSP即TMS320C6201的并行体系结构完成了对人脸识别算法的实时实现,通过这种多处理器并行系统,对CF2C算法进行合理的任务分派,系统达到了较好的实时性效果和较高的识别效率,同时系统通过ARM7最小系统,完成一些外围的控制功能,使DSP可以专注于信号处理。A couple of DSP, that is a parallel architecture consisting of two TMS320C6201, are used to guarantee the face recognition algorithm’s real-time implementation. With carefully designed task dispatch in this parallel system, better real-time effect and higher recognition efficiency can be achieved. Simultaneously, a simplified ARM7 module is included which is responsible for the peripheral equipment's control, thus the parallel DSP modules can focus on signal procession fully.

关 键 词：人脸识别 EFM 并行体系 DSP ARM7 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]