分析了DM642 的芯片和外围接口,提出了基于TI 公司的DSP 芯片- TMS320DM642 的人脸检测追踪方法。首先采用了YUV 模型在视频流中进行检测,得到肤色图像,然后求出各连通区域的外接矩形,通过脸部三角形的判断进行人脸确认。第一帧过后,采用了卡尔曼滤波进行了人脸追踪,减少了检测范围,提高了速度。关键词:DM642;人脸检测;人脸跟踪;卡尔曼滤波Abstract: The chip of DM642 and periphery circuit is analysed. A face tracing method is proposed, which based on DSP chip –TMS320DM642 of Ti. First of all, in order to local the face in videofrequency stream, we make use of YUV skin model. and then get the external rectangle of connected area. we lastly confirm the face by judged whether the triangle existed. After the first frame, we trace the face with KALMAN filter, reducing area and enhancing the speed.Key words: DM642 ;face detecting; face tracking; KALMAN filter
猜您喜欢
推荐内容
开源项目推荐 更多
热门活动
热门器件
用户搜过
随便看看
热门下载
评论