嵌入式3D-DWT-SPIHT壓縮算法的優(yōu)化和實(shí)現(xiàn).pdf

1、V9336JoL旦欠擘學(xué)校代碼：10246學(xué)號(hào)：032021024碩士學(xué)位論文嵌入式3DDWTSPIHT壓縮算法的優(yōu)化和實(shí)現(xiàn)院專姓指導(dǎo)教師：塑選熬攫完成日期：2QQ互箜5目15日AbstractTheblock—basedDiscreteCosineTransfolTn(DCTliswidelyusedinvideocodingstandardsavailabletoday，suchasITU’sH261，H263，H264andISO

2、／IEC’sMPEG一1，MPEG2，MPEG一4RecentresearchinimageandvideocodingalgorithmshasrevealedthatagreatercompressioncanbeachievedforthesamevisualqualityiftheblockbasedDCTapproachisreplacedbyaDiscreteWaveletTransform(DWT)Consequently

3、thelossycompressionalgorithmsbasedonreductionofspatialtemporalredundanciesusingDWTarousetheinterestofpeopleSincethewavelettransformhasmadeanunprecedentedhighperformanceforimagecompressionwithlowcomplexitytheMPEG21，thepur

4、poseofwhichistodefineanewscalablevideocodingtechnologywithhighcompressionperformance，isconsideringusingthe3DwavelettransformThe2Dzero—treecoding(EZW)developedbySaidandPearlmanhasbeenextendedtothreedimensions(3DEZW)byChen

5、andPearlmanIthasbeenshowntobeaneffectiveandcomputationallysinaplevideo—codingsystemwithoutanymotioncompensationandwithexcellentnumericalandvisualresultsKimandPearlman1aterdevelopedthe3Dsetpartitioninginthehierarchicaltre

6、es(3DSPIHDcodingalgorithmtoimprovethe3DEZWsystemfurtherItisunderstoodthatthe3DSPIHTisfastandefficientinvideocodingsincendoesnotrequirethemotionestimationandcompensationThecurrent3DSPlHTcodinghoweverrequirestheinformation

7、indicatingzerotreestructuresandstoringthreetemporarylistswhosesizewillquicklyexpandatapowerof2Sinceitiscostandtimeconsumingtousealargeamountofmemoryspacetostoretheselists，the3DSPIHTrequirementforbigmemoryhasbecomeacrucia

8、ldrawbackforitsreal—timeapplicationsInordertoimplementrealtimevideoencodinganddecoding，replacingtheoriginal3D—SPIHTalgorithm，anewembeddedvideocodecalgorithmnamed‘‘3DradialsubtreepartitioningSPIHT(3DRSPSPIHT)’ispresentedi

9、nthispaperWepartitionthewaveletimagecoefficientsintoanewstructure，ie，，3Dradialspatiotemporaltreestructure，whichcanbeencodedanddecodedindependentlybythesubtreebased3DSPIHTalgorithmInthiswayeachtemporarylistusedintheproces

10、sofsubtreebased3DSPIHTalgorithmcanberefreshedsothatonlysmallmemoryisrequiredfortheproposedalgorithmFurthermore，anextrabenefitearnedistheparallelizationrealizationofthecompressionanddecompressionalgorithmsOurvideocodecalg

11、orithmalsousesthe‘TruncateThresholdEstimation’and‘AdaptiveStreamDistribution’methodstooptimizeitsperformanceExperimentalresultsshowthatourapproachcandrasticallyreducethememoryrequirementto10％requiredbythe3DSPlHTalgorithm

12、requirementwhereasthedegradationiswithin03dBFurthermoreour3DRSP—SPIHTme出odisusedinhyperspectrumremotesensingimagecompression，whichalsoobtainedgoodperformanceKeywords：videocompression，setpartitionedinhierarchytrees(SPIHT)