预览加载中，请您耐心等待几秒...

基于OpenCV的物体定位与捕捉系统设计.docx

预览
1/2
2/2

在线预览结束，喜欢就下载吧，查找使用更方便

下载提示 文本预览

如果您无法下载资料，请参考说明：

1、部分资料下载需要金币，请确保您的账户上有足够的金币

2、已购买过的文档，再次下载不重复扣费

3、资料包下载后请先用软件解压，在使用对应软件打开

基于OpenCV的物体定位与捕捉系统设计 Introduction: Objectdetectionandcapturesystemhassignificantimportanceinvariousindustrial,commercialandmilitaryapplications.Inrecentyears,withtheadventofComputerVisionandMachineLearningtechnologies,thedevelopmentofsuchsystemshasbecomeevenmoreefficientandaccurate.OneofthemostcommonlyusedopensourcelibrariesforcomputervisionisOpenCV.Inthispaper,wepresentasystemtolocateandcaptureobjectsusingOpenCV. Methodology: Theproposedsystemconsistsoftwomaincomponents,namely,objectdetectionandobjectcapture.Weusedacameratocapturevideoframesandprocesstheminreal-timetodetectobjects.ThesystemwasdevelopedusingthePythonprogramminglanguageandOpenCVlibrary. Objectdetection: ObjectdetectionwasperformedthroughapopularobjectdetectionalgorithmcalledHaarClassifier.Weusedpre-trainedHaarCascadeclassifierstodetecttheobjectsfromthevideofeed.HaarCascadeclassifiersaretrainedtodetectobjectsofdifferentsizesandorientationsfromimagesbyusingpositiveandnegativeexamples. Inthissystem,wetrainedtheclassifiertodetecttheobjectswewantedtocapture.Weusedatotalof600positiveimagesand400negativeimagestotraintheclassifier.Positiveimagescontainedtheobjectofinterest,andnegativeimagescontainedrandomimages. Objectcapture: Afterdetectingtheobjectinreal-time,weusedtheServomotoralongwiththeArduinoboardtocapturetheobject.Servomotormotorswereusedtocontrolthehorizontalandverticalaxisoftheroboticarmthatholdsthecapturemechanism.Weusedasolenoidvalveforcapturingobjects. Results: Theproposedsystemwasimplementedandtestedinacontrolledenvironment.Theobjectsusedfortestingwereballsofdifferentcolorsandsizes.Thesystemwasabletodetecteachballaccuratelyandcaptureitusingtheroboticarmandsolenoidvalve. Conclusion: TheproposedsystemdemonstratedtheeffectivenessofusingOpenCVforobjectdetectionandcapture.TheuseofHaarCascadeclassifiersfordetectionandservomotorsalongwiththeArduinoboardforcaptureshowsgreatpotentialforfutureworkinthefieldofrobotics.Thissystemcanbefurtherimprovedandscaledforvariousindustrialandcommercialapplications.