基于wsn的视频与磁敏传感器道路车辆检测及其融合分析-road vehicle detection and fusion analysis of video and magnetic sensor based on wsn.docxVIP

低了算法实际运行时间，使算法更节能以适应无线传感网络节点能量有限的要求。仿真实验结果表明此磁敏车辆检测算法的准确较高。最后，本文提出基于这两种传感器融合的解决方案，讨论了其中传感器空间分布、时间同步、数据发布格式与发布方式等问题。重点讨论了用DS论据理论解决融合数据的相关与数据融合问题，并抽取磁敏信号的特征距离和视频信号的检测矩形框长度作为融合特征，进行多车道多类传感器融合，可以解决多车道大车覆盖小车、车队漏检的问题。研究无线传感网环境下的视频与磁敏传感器道路车辆检测及其融合在国内还处于起步阶段，本文提出了基于DS融合的解决方案，但还有很多问题有待进一步研究，如：适于道路车辆检测的无线传感网络内部的时间同步与网络层协议；如何选择道路车辆检测中的多传感器融合特征以提高融合的准确率等。关键词：跟踪车辆检测，无线传感网络，帧差法，磁敏传感器，多传感器融合，DS论据WSN BASED ROADVEHICLE DETECTIONAND FUSION STUDYUSINGVIDEOAND MAGNETIC SENSORABSTRACTTheissueofvehicledetectionisthebasicandkeyprobleminIntelligentTrafficSystemTechnology,assomebasictrafficparametersthatareacquiredbyvehicledetection,suchastrafficflow,averagevehiclespeedandroadduty-cyclingrateareimportantfoundationofimplementationofhigh-leveltrafficmanagementapplicationlikerouteguidanceandtrafficcontrol.Traditionalvehicledetectionmethodsuseonlyonetypeofsensor.However,constrainedbythecharacteristicofsensor,therearealwayssomedisadvantagesforsingle-type-sensormethodtodovehicledetection.Asthedevelopmentofwirelesssensornetworktechnology,peoplehavenoticedthatvehicledetectionproblemcanbesolvedbyusingwirelessnetworkwhichisself-organizedandconstructedbynodescontainingvarioustypesofsensor.Inthisway,systemcostwillbecutdownandsystemrobustnesswillbeincreased.Moreover,wecanobtainmoreaccuratetrafficparametersby usingmulti-sensormethods.Thispaperisfocusonwirelesssensornetworkenvironmentbasedvideomagneticsensorvehicledetectionalgorithmandmulti-sensordatafusion.Asvideosensorissensitivetochangeoflightnesswhilemagneticsensorisnot,magneticsensorcanbeusedtocomplementvideosensorinvehicledetectionwhenvideoisdisabledinday-nighttransformationandcomplexweathersituation.Animprovedvideodetectionalgorithmispresentedinthisthesis.Thealgorithmuseedgeinformationtoreducenoisepointswhichwascausedbyvideoshaking,soastoincreasetherobustnessofalgorithm,anduseROIarea-shrinkingmethodtoincreaseefficiency.ThisAlgorithmhaslowcalculationcomplexityandgoodreal-timeperformance,andhasgooddetectionresultinmulti-roadwaysituation.Theexperiment