基于DSP的视频人脸监控系统设计与实现

摘要

针对目前人脸监控系统存在不能及时决策、盲目跟踪的问题,提出综合人脸检测、识别、跟踪与控制的智能监控系统,并在TMS320DM642芯片上实现；使用Adaboost级联多层分类器检测人脸,基于Gabor小波图像相似度匹配识别人脸,对识别出的感兴趣人脸用Camshift算法实现跟踪,再通过串口驱动云台带动摄像头旋转,使感兴趣人脸保持在视场内并且在模拟显示器上显示;实验结果表明,本系统满足实时视频监控要求,对人脸旋转、遮挡、光照鲁棒性高;本系统可以推广至视频会议、视频聊天、可视电话等领域,有良好的应用前景.%In view of the current monitoring system can not be timely decision and blind tracking human faces, proposed an intelligent monitoring system, based on face detection, identification, tracking and control of intelligent monitoring systems, and its implementation in the TMS320DM642 chip. It applied AdaBoost algorithm to detect human face, face recognition based on Gabor wavelet. Camshift algorithm is used to track identified face. Through the serial correspond with PTZ driven camera rotation, let interested people face always in the scope of the monitoring and display on the monitor. Experimental results show that, this system can meet real-time requirement, and have good robustness of rotation, occlusion, and illumination. This system can be extended to areas such as video conferencing, video chat, video phone, there are good prospects.