中文摘要：

针对视频监控系统人工监视容易产生疏漏的问题,研究在铁路场景下基于智能视频监控图像序列处理和分析的识别物体入侵行为检测技术,给出铁路入侵物体检测、定位与跟踪的适用算法,实现基于轨迹点行为模型的入侵行为分析与理解。设计和开发了基于OpenCV的铁路入侵检测实验平台,在此平台上实现背景建模、运动检测与跟踪等项功能,初步实现了对入侵行为的识别、分析以及对危险行为的报警。

英文摘要：

Focusing on the problem easily leading to the oversights in the manual video surveillance system, this paper studies the invasion detection technology in the railway scene based on the image sequence processing and analysis, which gives the algorithms of the invasion object detecting, locating and tracking, and the approaches for behavior analysis and interpretation with the point trajectory behavior model. With the OpenCV tools, an experiment platform for the railway invasion detection is designed and developed. This platform has accomplished the background modeling, motion detecting and tracking etc.. And it can initially realize the identification and analysis of the invasion behavior, and the warning for the dangerous behavior.