使用加权4D-DCT基础重建压缩采样光场.pdfVIP

RECONSTRUCTIONOFCOMPRESSIVELYSAMPLEDLIGHTFIELDS

USINGAWEIGHTED4D-DCTBASIS

YusukeMIYAGI,KeitaTAKAHASHI,MehrdadPANAHPOURTEHRANI,ToshiakiFUJII

GraduateSchoolofEngineering,NagoyaUniversity

DepartmentofElectricalEngineeringandComputerScience

Furo-cho,Chikusa,Nagoya,464-8603,Japan

individualpositionsanddirectionsthroughasinglecamera.

Thecodedaperture/masktechniqueenabustocapturelightHowever,suchacamerahasatrade-offweenpositional

fielddatainacompressivewaythroughasinglecamera.Aanddirectionalresolutions.Thelastwayisusingacoded

pixelvaluerecordedbysuchacameraisasummationoftheaperture/maskwithasinglecamera[13,14,15,16,17,18],

lightraysthatpassthoughdifferentpositionsonthecodedwhichisfocusedoninthispaper.

aperture/mask.ThetargetlightfieldcanbereconstructedAcodedaperture/maskcameracancapturealightfield

fromtherecordedpixelvaluesbyusingpriorinformationofwithoutdecreasingtheresolution,butalargernumberof

thelightfieldsignal.Aspriorinformation,adictionary(lightshots(numberofcapturedimages)isrequiredforhigher

fieldatoms),whichwaslearnedfromtrainingdatasets,wasqualityacquisition.In[16,17,18],thenumberofshotswas

usedinthecurrentstateoftheart.Meanwhile,itwasre-significantlyreducedbyusingpriorinfor