视频监控中的人体检测与跟踪技术研究-信息与通信工程专业论文.docx

下载文档 降价啦

1
0
约5万字
约 63页
2018-05-05 发布于上海
举报
版权申诉
保障服务

视频监控中的人体检测与跟踪技术研究-信息与通信工程专业论文.docx

1、本文档共63页，可阅读全部内容。
2、原创力文档（book118）网站文档一经付费（服务费），不意味着购买了该文档的版权，仅供个人/单位学习、研究之用，不得用于商业用途，未经授权，严禁复制、发行、汇编、翻译或者网络传播等，侵权必究。
3、本站所有内容均由合作方或网友上传，本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺！文档内容仅供研究参考，付费前请自行鉴别。如您付费，意味着您自己接受本站规则且自行承担风险，本站不退款、不进行额外附加服务；查看《如何避免下载的几个坑》。如果您已付费下载过本站文档，您可以点击这里二次下载。
4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等，请点击“版权申诉”（推荐），也可以打举报电话：400-050-0827(电话支持时间：9:00-18:30)。

视频监控中的人体检测与跟踪技术研究-信息与通信工程专业论文

AbstractPedestriandetectionandtrackinginsurveillanceisanimportantresearchissueofcomputervisionandpa忧ernrecognition，currently.Duetoincreasingrequirementofsecurityinpublicplaces，moreandmorehumanandfinancialresourcesfrommanywell-known companiesandresearchinstitutionspaγfortheresearchoftheintelligentmonitoringsystem.Thistopiccanbepredictedtohavebroadprospectsandtobringabouttremendoussocialbenefitsinfuture.This paperconstructedavideosequencebasedpedestriandetectionandtracking system，whichincludedfourpa同s:movingobjectdetection，。问ecttrackinginsinglecamera，facedetectionandpedestriantrackingincameranetworks.First，movementinformationwasusedtosegmentforeground regions，whichwouldbetrackedinvideosequencesfromonecamera.Theninordertodeterminewhetherforegroundregionswerehuman，facedetectionwasadopted.Finally，integrationofhumanbodyandfaceareafrommultiplecamerascouldtrackpedestrianinthenetworkofcameras.Inmovingobjectdetection，Backgroundsubtractionalgorithmwasresearchedforboththestaticanddynamicscenes.SingleGaussianmodelforindoorenvironment(staticscenes)andMixtureofGaussianmodelforoutdoorenvironment(dynamicscenes)effectivelycompletedthetaskofmovingobjectdetection.Colormodelwasadaptedtosolvetheproblemofshadow.Forsinglecamera，threetrackingalgorithmswerecompared:trackingbasedonconnectedcomponent，trackingbasedonMeanSh的andtrackingbasedonParticlefilter.A fusion algorithm was proposed:absence ofobject occlusion，the rapidalgorithm based on connected component was adopted; appearing occlusion，particlefilterwithMeanSh的wasused，whichgreatlyreducedtheprocessingtimeandimprovedtrackingaccuracy.Inordertodeterminewhetherthetracked targetstobeahumanandpositionthefaceregion forfacerecognition，the旬cedetectionalgorithmwasresearched.Differentfromthetraditionalfacedetection，duetothesmallhumanbodyinvideo surveillance，thispaperdetectedfaceusingpa目ialinformationofupperbod机includingface，shoulderandhead. HarrfeaturesfromtheupperbodywereextractedandAdaboostalgorithmwasusedforthetrainingandtesting.Thispapersimplif