智能监控系统中的区域入侵检测与跟踪摘要区域入侵检测与跟踪是智能视频监控系统的主要应用之一，常用于银行、铁路、机场、博物馆等重要民用设施的安全防范监控，此外，在军事上也有着广阔的应用前景，如对边海防、哨所、军队仓库等敏感区域的实时监控。本文对智能监控系统中的区域入侵检测与跟踪的基本算法进行了研究，并设计实现了一个基于Visual C++ 6.0平台的实验系统。首先介绍了智能视频分析技术的研究意义、发展状况和趋势，然后详细介绍了目标检测与提取的常用算法，并对基于外接矩形和颜色特征的跟踪方法进行了研究。在对已有算法研究的基础上，提出了一种基于像素级和帧级相结合的背景模型更新与去噪处理方法，有效地解决了因静止运动目标带来的长时间检测错误，同时消除了大面积随机噪声的干扰。最后，给出了该系统在不同室外场景下的区域入侵检测与跟踪的实验结果和分析。
关键词：区域入侵，目标检测，目标跟踪，背景模型，智能监控
egion Intrusion Detection and Tracking in Intelligent Video Surveillance System
Abstract
Region intrusion detection and tracking is a major application of intelligent video surveillance system, which is usually used in security surveillance and monitoring of important civilian facilities such as banks, railway stations, airports, museums and also has a wide application prospect in military security.
This paper studies the basic algorithms for region intrusion detection and tracking in intelligent video surveillance system, and develops an experiment system based on Visual C++ 6.0 platform. After reviewing the research significance of intelligent video surveillance technologies and its current development and directions for future research, this paper describes the common-used algorithms for object detection and segmentation in details, and then put emphasis on discussing tracking methods based on bounding box and color feature. On the basis of studying existing algorithms, a novel approach for background model updating and noise cleaning by combining pixel-level and frame-level is proposed in this paper, which is able to solve false positives caused by static moving objects effectively and eliminate the random noises with large area. Finally, the test results of this system under different outdoor environments are shown and analyzed.
Keywords: Intrusion Detection, Object Detection, Object Tracking, Background Model, Intelligent Video Surveillance
*若需了解更多与协助请咨询↓→[电脑QQ][手机QQ]【数据协助】
