运动目标检测中阴影去除算法的研究与实现毕业设计正文

JIANGSUUNIVERSITY本科生毕业论文运动目标检测中阴影去除算法的研究与实现ResearchandrealizationoftheshadowremovingalgorithmforMovingobjectdetection学院名称：计算机科学与通信工程学院专业班级：通信工程0602班学生姓名：汪雅洁指导教师姓名：宋雪桦指导教师职称：副教授2010年6月江苏大学本科毕业论文I运动目标检测中阴影去除算法的研究与实现专业班级：通信工程0602班学生姓名：汪雅洁指导教师：宋雪桦职称：副教授摘要随着计算机视觉技术、电子技术、通信技术的发展，智能视频监控系统作为安全防卫的一种重要手段正在越来越受到人们的重视。由于智能视频监控系统具有监控能力强、安全隐患少、节省人力物力资源的优点。因此，在交通、银行、宾馆、商场等重要场所的监控中有广泛的应用前景。本文首先综合介绍了智能监控系统的发展历史和现状，然后对静止摄像机监控下的运动目标检测、阴影的检测和去除等关键技术进行了比较深入的研究。运动目标检测作为智能视频监控系统中视频处理的第一步，具有非常重要的地位。本文首先对目前运动目标检测方法进行了概括，在详细研究了几种目标检测方法的基础上，确定了目标检测中较好的一种方法，即基于混合高斯模型的方法，用这个算法来提取运动目标。由于日照和灯光等外来因素的影响，造成了提取的运动前景中往往含有阴影。因此，运动目标的阴影检测与去除对于运动目标跟踪、分类和识别等后期处理都是一个关键性问题。由于阴影的存在，会给上述后期处理带来干扰甚至失败。为了去除目标前景的阴影，本文首先分析了阴影产生的机理，了解阴影的特征和人类的视觉特征，针对这些特征以及总结和分类目前己有的各类阴影检测算法的基础上，提出了一种基于RGB颜色模型的阴影检测算法。通过实验对本文的算法进行了验证，证明了该算法能够很好地检测出运动目标的阴影以及将阴影去除，而且易于实现。关键词：视频监控；运动目标检测；混合高斯模型；RGB颜色模型；阴影去除江苏大学本科毕业论文IIResearchandrealizationoftheshadowremovingalgorithmforMovingobjectdetectionAbstractDevelopmentofthecomputervisiontechnology,theelectronicandthecommunicationtechnology,hasmadetheintelligentvisualsurveillancesystemanincreasinglyimportantsafedefenseway.Becauseithasadvantagesofhigherqualityandlessneedofinvestment.Soithascheerfulprospectintheapplicationsofsurveillancefortraffic,bank,hotel,shopping,etc.Bothitshistoryandcurrentsituationissummarizedhere,then,aresearchwasmadeforthekeytechnologyofthesegmentationofmovingobjectsandthedetectionandremovalofshadows.Astheinitialstageinthevisualdataprocessing,movingobjectdetectionisakeypoint.Aftercarefullystudyofmovingobjectdetectionmethodsusedpresently,amorereliablealgorithmisdetermined,thatis,themixedGaussmodel.Itwasadoptedtodetectmovingobjects.Asexternalfactorssuchassunlightandlightingeffects，resultinginextractionofmovingforegroundoftencontainshadow.So,shadowsdetectionandeliminationofmovingobjectsisessentialtothepost-processingsuchasobjectstracking,classificationandrecognition.Theexistenceofshadowwillallowtheabove-mentionedpost-processingtofail.Inordertoremovetheshadowofobjectforeground,thispaperfirstanalyzethemechanismoftheshadowproduced,understandthecharacteristicsoftheshadowandthehumanvisualcharacteristics,then,amethodofshadowdetectionbasedontheRGBcolormodelisproposedonthebasisofthesecharacteriticsandthesummaryandanalysisforvariousshadowdetection.Wehaveconductedmanyexperimentstoverifytheproposedapproach.Theresultsshowthatthealgorithmcandetectmovingtargetstoremovetheshadow,andeasytoimplement.KeywordsVisualSurveillance;MovingObjectDetection;MixedGaussianModel;RGBcolormodel;ShadowRemoval江苏大学本科毕业论文III目录第一章绪论...................................................................................................................................11.1引言...................................................................................................................................11.2视频监控系统的发展和现状...........................................................................................11.3本课题研究的目的及意义...............................................................................................21.4课题主要研究工作及工作安排.......................................................................................3第二章基础理论...........................................................................................................................52.1引言...................................................................................................................................52.2颜色模型...........................................................................................................................52.2.1颜色模型的分类.....................................................................................................52.2.2RGB颜色模型.........................................................................................................52.2.3HSV颜色模型.........................................................................................................62.3数学形态学.......................................................................................................................72.3.1基本思想.................................................................................................................82.3.2基本运算.................................................................................................................8第三章运动目标的检测.............................................................................................................103.1引言.................................................................................................................................103.2运动目标检测算法概述.................................................................................................103.2.1光流法...................................................................................................................103.2.2相邻帧差法...........................................................................................................103.2.3背景差法...............................................................................................................113.3基于RGB颜色空间的混合高斯模型...