摘要
针对现有视频监控系统的缺陷,提出一种新的基于人脸检测与跟踪的智能监控系统。利用对称差分算法,自动检测场景中的运动区域,限制搜索范围;然后利用BP神经网络对肤色进行识别,获得候选人脸区域,该方法比固定阈值肤色检测方法具有更强的环境适应能力;经过人脸验证,最终定位图像中的人脸;对检测出的人脸,提出了新的基于肤色信息和维护运动人脸缓冲池的方法,主动跟踪目标人脸。依据检测出的人脸信息和当前的日期、时间,建立相应的监控信息标注数据库,以供后期查询。实验表明,该系统能够实时可靠地检测、跟踪运动人脸,满足特定的监控要求。
Aiming at ractifying the shortcomings of traditional video surveillance system, an intelligent surveillance system that can automatically detect and track human faces in the scene is presented. Symmetrical frame difference is used to acquire the area of motion and the skin-color segmentation algorithm based on BP neural network is used to extract the face candidates. Then, the candidate face regions are verified with the knowledge of human faces. A recorded face buffer is maintained to track moving faces in the scene. The captured faces and event of interest are used to generate video indexing database. Experimental results show that the intelligent surveillance system can detect and track the faces rapidly and accurately in the scene.
出处
《北京理工大学学报》
EI
CAS
CSCD
北大核心
2004年第11期966-970,共5页
Transactions of Beijing Institute of Technology
基金
国家自然科学基金资助项目(69973007)
关键词
视频监控
人脸检测与跟踪
对称差分
BP神经网络
video surveillance
face detection and tracking
symmetrical frame difference
BP neural network
