Abstract: Intelligent surveillance is one of the hot fields in computer vision.This paper presents an intelligent detecting and tracking surveillant system based on video analysis，which is suitable for indoor environment.Firstly，a statistical model-based detection algorithm is used to find moving object.Then，the moving object is roughly tracked using meanshift.Meanwhile，to improve the deficiency that the scale of meanshift is not changeable，an adaptive scale update algorithm based on boundary force is presented.Finally，an intelligent video surveillant platform is designed and constructed with DM642 as the core.The experimental results show that the system can detect and track the moving object in indoor scene rapidly and effectively.

Key words: intelligent surveillance, DM642, Meanshift

摘要： 智能监控是当前计算机视觉研究的热点领域之一。针对室内监控的具体特点，实现了一种基于视频分析的智能检测跟踪监控系统。利用基于统计模型的目标检测算法提取运动目标；然后结合Meanshift算法对目标进行粗跟踪；最后针对Meanshift算法无法实时改变跟踪窗大小的缺陷，提出边界力调整算法以自适应更新跟踪窗窗宽。并以DM642数字图像处理DSP为核心，设计并搭建了智能视频监控平台。实验表明，该系统可以实时有效地检测、跟踪室内运动目标。

关键词: 智能监控, DM642, Meanshift