基于Bandelets的遙感圖像壓縮技術(shù)研究.pdf

1、西安電子科技大學(xué)碩士學(xué)位論文基于Belets的遙感圖像壓縮技術(shù)研究姓名：孟會曉申請學(xué)位級別：碩士專業(yè)：電路與系統(tǒng)指導(dǎo)教師：楊淑媛20100101AbstractAbstractWithrapiddevelopmentofremotesensingtechnology，theremotesensingimagesarewidelyusedinthefieldoflandresourcesinvestigationThelimitedcha

2、nnelcapacitycallnotmeetthedemandoftransmittingthegreatamountofremotesensingdataMultiscalegeometryanalysisisanewtoolthatissuperiortowaveletBecauseofthecharacteristicofanisotropyandmulti—directional，multiscalegeometryanaly

3、sistoolsCanrepresenthigh—dimensionalsingularitiesmoreeffectivelyBandeletsarethetypicaladaptivemultiscalegeometryanalysistoolsTheycarltakeadvantagesofintrinsicgeometricregularityofimagestructuresandobtaintheoptimalreprese

4、ntationofunderlyingfunctionsButtherealizationoffirstgenerationbandeletsisverycomplicatedThesecondgenerationbandeletsCanavoidtheadaptivequadtreepartitiontheexhaustivesearchingfortheoptimalgeometricflowsSOenablesafastandef

5、ficientcompresslonThispaperfocusesontheBandeletsapplicationonremotesensingimagecompressionAimattwokindsofremotesensingimages，SARimagesandhyperspectralremotesensingimages，threealgorithmsforremotesensingimagecompressionbas

6、edonBandeletsareproposedSummarythemainworkasfollows(1)Inordertomakeagooduseofinformationineachhighfrequencysub—bandtopreservethedetailedandtextureinformation，weproposedamethodbasedonWaveletPacketBandeletsforSARimagesThec

7、oreofthismethodstillusesthefixedpartitionsize，butadoptWaveletPacketdecompositionformultiresolutionanalysisbeforeperformingBandeletsThesimulationresultsshowthatthesuperiorityofourmethodtothesecondgenerationBandeletsandBan

8、delets、Ⅳitllthefixedpartitionsizeinperformance，謝mahigherqualityandabeRervisualeffectofthereconstructedimage。I’lliSiSalleffectivemethodforSARimagecompression(2)BasedoncharacteristicsofsparsityofBandeletscoefficients，andap

9、plythemultiscaleBandeletswithadaptivequadtreepartitiontoencodethemBecauseBandeletSCanprovideanefficientrepresentationofgeometricstructuresexistedinimages，theBandeletsbasedmethodCancaptureabundanttexturesinSARimagesAndthe

10、coefficientsdecomposedbyBandeletshaveasimilarstructurewiththatofwaveletcoefficientsEmbeddedBlockCodingwitllOptimalTruncation(EBCOT)codingalgorithmisemployedinOurproposedmethodExperimentalresultsshowthatOUrmethodoutperfor