摘要
Super-Resolution(SR)techniquemeanstoreconstructHigh-Resolution(HR)imagesfromasequenceofLow-Resolution(LR)observations,whichhasbeenagreatfocusforcompressedvideo.BasedonthetheoryofProjectionOntoConvexSet(POCS),thispaperconstructsQuantizationConstraintSet(QCS)usingthequantizationinformationextractedfromthevideobitstream.BycombiningthestatisticalpropertiesofimageandtheHumanVisualSystem(HVS),anovelAdaptiveQuantizationConstraintSet(AQCS)isproposed.SimulationresultsshowthatAQCS-basedSRal-gorithmconvergesatafastrateandobtainsbetterperformanceinbothobjectiveandsubjectivequality,whichisapplicableforcompressedvideo.
出版日期
2007年03月13日(中国期刊网平台首次上网日期,不代表论文的发表时间)