基于DSP的实时人脸检测系统

Real Time Human Face Detection System Based On DSP

阐述了硬件设计思想、软件开发流程及人脸检测算法的实现。首先以TMS320DM642建立实时图像处理的硬件系统,通过CCD获取视频序列,对当前图像进行降噪、图像增强以及边缘检测的预处理,再利用基于肤色和差分帧相结合的算法对视频序列进行检测、跟踪人脸。本系统所有算法均使用C和汇编语言混合编程并在TIDM642EVM上开发实现。实验结果表明,本系统的算法简单、快速、鲁棒性强。

The designing ideas of the hardware and developing flow of the software and the algorithm of human face detection are expounded. At first setting up the real-time video processing system of the hardware using TMS320DM642, obtaining video sequence with CCD, carrying on noise reduction, image strengthen and edge detection that pretreatment measure to current image, human face detection and track are combined using based-on skin-color and difference frame algorithm. All algorithms of the system use C and assembler language mix-program and develop and realize on TI DM642EVM. Experiments show that the algorithm is simple, fast and robust.