视频中目标轨迹提取算法设计与实现.doc

摘 要题题 目目 视频中目标轨迹提取算法 设计与设计 摘 要视频中运动目标轨迹的提取与分析是智能监控视频分析的重要基础目前世界上覆盖着数以千万计的视频监控摄像头，这些监控摄像头在航天、军事、交通、安防等各个领域均发挥着重要的作用当需要查找某个关键事件时，如果使用传统人为快进快退的方式，那么将会耗费大量的人力而基于提取视频目标的视频摘要技术的实现将大大提高关键事件查找的效率和准确率本文在对基于运动对象的视频摘要技术进行研究的基础上，论述了基于运动对象的视频摘要系统的基本框架，给出了基于混合高斯模型的背景建模，该算法解决了光线亮度变化、阴影、遮挡以及非静态背景问题，精确的检测出运动目标论述了基于均值漂移和粒子滤波的跟踪算法，实现了多目标、目标碰撞情况下的轨迹提取最后，在轨迹后处理模块中，使用线性插值的方法对运动目标的轨迹进行插值处理，有效的解决了运动目标轨迹抖动漂移的问题本文采用基于 OpenCV 结合 VS2012 实现了视频中目标轨迹的提取关键词：视频摘要 目标检测跟踪 混合高斯背景模型 轨迹提取 轨迹处理ABSTRACTABSTRACTExtraction and analysis of the video moving target trajectory is an important foundation for intelligent surveillance video analysis.The world is covered with tens of millions of video surveillance WebCam, these surveillance WebCam play an improtant role in aerospace, military, transportation, security and other fields. When you need to find a key event, If you are using the traditional man-made fast-forward and rewind the way, it will spend a lot of manpower. While if you using Moving Object Trajectory based video abstract, you can quickly check over a certain period of time.In this paper, we discussed moving objects based on the basic framework of video synopsis after we researched moving objects based video synopsis. we given a moving objects detection based background model of Mixture of Gaussian. the algorithm solves the light intensity changes, shadows, occlusion and the problem of non-static background, accurate detection of moving targets.Finally,trajectory post-processing module,we using a linear interpolation method for processing the trajectory of the moving object, and effective solution to the problem of moving target trajectory drift. Our paper will using VS2012 with OpenCV for moving object extraction.Keywords: Video Synopsis Moving Detection and Tracking Mixture Of Gaussian Background Model Trajectory Extraction Trajectory Processing目 录i目 录第一章第一章 绪论绪论...................................................................................................................11.1 引言...................................................................................................................11.2 国内外研究现状...............................................................................................21.2.1 视频摘要.................................................................................................31.2.2 运动目标检测.........................................................................................41.2.3 运动目标跟踪.........................................................................................51.3 本文的研究内容...............................................................................................7第二章第二章 基于运动对象的视频摘要系统基于运动对象的视频摘要系统.......................................................................92.1 视频摘要技术分类...........................................................................................92.1.1 静态视频摘要.......................................................................................102.1.2 动态视频摘要.......................................................................................112.2 基于运动对象视频摘要.................................................................................142.2.1 视频摘要系统基本框架.......................................................................142.2.2 运动目标检测跟踪...............................................................................152.2.3 运动目标轨迹提取及后处理...............................................................15第三章第三章 运动目标检测运动目标检测.................................................................................................173.1 运动目标检测方法概述.................................................................................173.2 常用的运动目标检测.....................................................................................173.2.1 帧间差分法...........................................................................................173.2.2 光流场法...............................................................................................183.2.3 背景建模法...........................................................................................193.3 背景建模方法.................................................................................................203.3.1 均值法建模...........................................................................................203.3.2 中值法建模...........................................................................................203.3.3 单高斯模型...........................................................................................213.3.4 混合高斯模型.......................................................................................223.4 实验结果.........................................................................................................23目 录ii第四章第四章 运动目标跟踪运动目标跟踪.................................................................................................254.1 常用运动目标跟踪方法.................................................................................254.1.1 基于特征匹配跟踪方法.......................。