112RobotVision-4-VisionSystems2012/9/14ROBOTVISIONHuang-NanHuang2012/9/142012/9/142012FallOutline4.1HumanVisualApparatus32012/9/1442012/9/141thevisualcortex2thelateralgeniculate(彎曲狀)bodies
HumanVisualApparatus5theiris
ModelforTechnicalVisionSystems3thevisualnerve
CameraTypes4theretina6thecurvatureofthelens•Photosensitivecellsontheretina:Conesareresponsibleforthedeterminationofthegrayvaluesandthreetypesofrodsfortherecognitionofthecolorsred,greenandblue(RGB).Theconeshaveahigherphotosensitivitythantherods.•Thecolorsignalsoftherodsarecompoundedafterbeinggeneratingandthentransmittedbythevisualnerve(3)andthelateralgeniculatebodies(2)tothevisualcortex(1).Thelateralgeniculatebodiesareresponsibleforstereovision.•ThesignalsarestoredinLuv-channels.TheL-channelinvolvesinformationonthebrightnessandtheuv-channelsonthecolors.4.2TheHumanVisualApparatusasLayersforVisualCortex(腦皮質）ModelforTechnicalVisionSystems52012/9/1462012/9/14
TheobjectiveisresponsibleforfocusingandcanbecomparedwiththelensandtheCCD1)arrayisorientedattheretinaAttentionControlAttentionControl72012/9/1482012/9/14
Theadvantageofthehumanvisualapparatusisthat
Krotkov[57]combinesastereovisionsystemandafocusingthevisualsearchorobjectdetectionfindsinstancesfromapproachandusesthevisualattentioncontroltoobtainaobjectsbyoftenusingattributeslikethecolor.partialreconstruction.
Inthefirststepthestereovisionsystemisusedtogeneratearathercoarsereconstruction.Pointsinthescenearedetermined
Ananalogousapproach:Theentiresceneisanalyzedbytheuseofthecoarsereconstruction,whicharethenusedforabyoperatorstodetectstoredobjectmodelsinmoreprecisereconstructionwiththefocusingapproach.knowledgebase.Ifthescenepossessesobjectsthat
ThesystemfromAbbottandAhuja[58]developsglobaldepthmapsbystepwiseexpansiontorepresentthescenesuccessivelyhavearesemblancetotheobjectmodels,butbythemorecompletely.Ifadepthmapisgenerated,thebordersoftheresolutionofthewholescenetheequalitycannotmapareexaminedtoobtai