基于视频分析的实时跟踪及入侵警报系统  

Real-time tracking and intrusion alarm system based on video analysis

作　　者：陈璋雯[1] 达飞鹏[1] 陆可[2] 

机构地区：[1]东南大学自动化学院,南京210096 [2]南京工业大学国际教育学院,南京211816

出　　处：《东南大学学报（自然科学版）》2012年第A01期51-54,共4页Journal of Southeast University：Natural Science Edition

基　　金：国家自然科学基金资助项目(51175081;61107001);江苏省自然科学基金资助项目(BK2010058)

摘　　要：提出了一种基于视频分析的实时跟踪及入侵警报算法并构建了基于此算法的实时系统.运用Otsu算法对视频图像进行预处理.通过对预处理后的视频图像进行详细分析,根据运动行人的二值图像特征,用阈值分割及连通域的融合等方法实现对非目标人体干扰的去除,同时,对运动行人区域进行有效提取及跟踪.当实时跟踪的行人越过事先确定的警报区域,则系统发出警报.所提方法运算复杂度小,鲁棒性高,运算速度快,构建的实际系统可有效实现行人跟踪及入侵警报,实验证明该系统精度满足实际需求,具有较好的实时性,可应于较多领域.Abstract： This paper presents a realtime tracking and intrusion alarm algorithm based on video anal ysis, and a realtime system is built based on the algorithm. Video images are preprocessed by the Otsu algorithm. The proposed method makes a detailed analysis of the video image after pretreat ment. The threshold segmentation and the connected domain fusion method are used to remove non target human interference according to the characteristics of the moving pedestrians in the binary im age. Meantime, the area of the moving pedestrians are effectively extracted and tracked. The system alerts when the realtime tracking pedestrian overgoes the predetermined alarm zone. The proposed method is of low computational complexity, high robustness and high computing speed. The con structed actual system can effectively achieve the pedestrian tracking and intrusion alarm. Experimen tal results show that the system meets the requirements of precision and has a better realtime per formance, and it can be used in more areas.

关 键 词：智能视觉跟踪 入侵警报 行人跟踪 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]