基于数字图像的信息隐藏技术研究

华中科技大学硕士学位论文基于数字图像的信息隐藏技术研究姓名：万宝平申请学位级别：硕士专业：计算机系统结构指导教师：鲁宏伟20060429ILSB256LSBHVSIIAbstractWiththedevelopmentofcomputer,networkandmultimediatechniques,digitalproducts,suchastext,image,audio,video,etc,havebeenincreasinglypopularizedandwidelyapplied.However,theissuesofcopyrightprotectionandinformationsecurityhavebecomemoreseriouslythaneverduetothefacilitiesofdigitalmultimedia,suchaseasilycopied,edited,storedandpromulgatedthroughnetworks.Intelligencepropertyprotectionofdigitalproductshasbeenpaidmoreattentionbyallgovermentsandhasbecometheoneofstrikingproblemstobesolvedurgentlyindigitalworld.informationhidingtechniqueisnewbutboominginthefieldofinformationsecurity.Informationhidingbasedondigitalimageisoneofthemostimportantresearchbranch.Throughanalyzingandsummarizingthebasicideaandtheoriesofimageinformationhiding,andanatomizinghumanvisualcharacteristicindetail,animprovedLSBthatbasedonBMPimageandthetheorythatdifferenthumaneyessensitivitiestodifferentcolorsinallkindsofmaskingcharacteristicsofthehumaneyesandthebrightnessformulawaspresented.Inthisinformationhidingalgorithm,256gradesimagewasusedinthepublicimage,anyinformationsuchasimageandtextcanbethoughtasbinarybitstreaminformation,onlyiftheextractercanregrouptheextractedinformationondemand.ThepurposeofthisimprovedLSBtheimprovedmethodthatachievedthebetterimperceptibilityandthemorecapacityofhidinginformationaswellasthemoreavailabilityofhidinginformation.Experimentalresultsshowedthattheimprovedmethodachievedthebetterimperceptibilityandthemorecapacityofhidinginformation.Accordingly,theconflictbetweenthetechnicalrequirementsforaninformationhidingsystemwasalleviated.Atthesametime,aconcreteimplementationapproachbasedonthevalueofnoisethateffecthumanvisualsystemwasproposed.KeywordsImageinformationhidingImagequalityevaluationHumanvisualsystemScramblingTwo-pixelblock111.1Internet()[1](Encryption)19965(FirstInternationalWorkshoponInformationHiding)----[2](InformationHiding)()2[3]801.2InternetMIT,Purdue,Columbia3GeorgeMasonErlangen-NurembergNECIBMAT&TBell[4]863[5]12/1[6]2[1]341.3(HVS)/HVSLSBLSB522.12.1.1SteganographyDemertusSteganography()6()2.1.2(InformationHiding)2.1[7,35](Steganography)(CopyrightMarking)72.22.22.2(1)(SecretMessage)(2)(CoverImage)(3)(StegoImage)(4)(StegoKey)(SecretData)(5)(Embedding)(6)(Extracting)(7)(Steganalysis)(CommunicationChannel)8()()2.3[844](1)(Imperceptibility)(Undetectability)(2)(Capacity)(3)(Robustness)(Immunity)2.42.4.1[9]9(1)(2)(3)2.4.22.4.2.1(RMSE)(PSNR)(,)fxy'(,)fxy(,)fxy(,)fxyxy,f'(,)fxy[40]e(x,y)='(,)fxy(,)fxy(2-1)MN1100['(,)(,)]MNxyfxyfxy--==-∑∑(2-2)(,)fxy)'(,)fxyRMSE112001['(,)(,)]MNxyRMSEfxyfxyMN--===-∑∑(2-3)10RMSE'(,)fxy(,)fxye(x,y)msSNR1120011200'(,)['(,)(,)]MNxyMNmsxyfxySNRfxyfxy--==--===-∑∑∑∑(2-4)mnsSNRSNR(dB)11001(,)MNxyffxyMN--===∑∑(2-5)1120011200[(,)]10lg['(,)(,)]MNxyMNxyfxyfSNRfxyfxy--==--==⎡⎤-⎢⎥⎢⎥=⎢⎥-⎢⎥⎣⎦∑∑∑∑(2-6)maxmax{(,),0,1,...,1,0,1,...,1}ffxyxMyN==-=-2max1120010lg['(,)(,)]MNxyMNfPSNRfxyfxy--==⎡⎤⎢⎥××⎢⎥=⎢⎥-⎢⎥⎣⎦∑∑(2-7)PSNRPSNR'ff2.4.2.2(NVF)11NVF[53]21(,)1(,)XNVFijijqs=+•q2(,)Xijs(,)ij(1,)ijM≤≤2(,)Xijs()2221(,)(,)(,)(21)LLXkLlLijxikjlxijLs=-=-=++-+∑∑221(,)(,)(21)LLkLlLxijxikjlL=-=-=+++∑∑(2L+1)×(2L+1)NVFqq2maxxDqs=2maxxsD∈50,100NVFNVF1NVF0HVSNVF(1)CSM×NΔX{},,,0;0mnmnXXCSmMnNΔ=Δ=-≤≤(2)xp×p11,0010,0ppijimjnmnMNMBMBXijppPP--++==⎧⎫==Δ≤≤⎨⎬×⎩⎭∑∑12(3)Sp×p11,0010,0ppijimjnmnMNLLSijppPP--++==⎧⎫==≤≤⎨⎬×⎩⎭∑∑ijLminijijKCL=K(4)ijMBminijijijMBCL0ijMB=ijMB(5)2-8×p11,0010,0PPijimjnmnMNMBNVFMBNVFNVFijPPPP--++==⎧⎫==≤≤⎨⎬×⎩⎭∑∑(6)0,0ijijijMNMaskMaskMBNVFMBijPP⎧⎫==×≤≤⎨⎬⎩⎭(7)(stego-imagequalityevaluationSIQE)210lgijijMaskSIQENMPP=⎢⎥⎢⎥×⎣⎦⎣⎦∑SIQESIQE[36]2.5(SpatialDomain)(TransformationDomain)132.5.1(LeastSignificantBits,LSB)[10,40,41]LSBBanderPatchwork[37]N(ia,ib)1;1,2,...,1iiiiaaiNbb=+⎧=⎨=-⎩PatchworkLSBJPEGFIRBanderTanaka14[5](blind)2.5.2DCT)[11,38](Wavelet)[12,39]DCT(1)(2)DCTDCkDCT{},1,2,...,iXxik=={},1,2,...,iWwik=={1},1,2,...,iiixxawik=+=(2-12)(a)k(3)IDCT(8×8)DCT(2-12)(1)(2)(LH2,HL2)h15m'(,,)fhfmb=(2-13)f'f,b(3)(2-13)mDCTDFT[13]DCTLL,LH,HL,HHLL;(N*N)DWTO(N),DCTO(N*IogN),DWTDCT(N)DWTDCTMPEG-4JPEG200(1)(2)16)(3)(4)[14]HVS(HumanVisualSystem)[37,42,43]HVSHVS173(HumanVisualSystem)(MaskingCharacteristics),HVS[15]/HVS3.13.1.1humanvisualsystem,HVS18[16]HVS[17]3.1.2(l)Weber[18]II+II≈I*0.02(2)[19](3)()()[20](4)19(Self-contrastMasking)(Neighborhoodmasking);(5)(6)(7)[21](8)[22][23](MaskingCharacteristics)3.2HVSHVSHVSHVS20HVS(Entropy)3.2.1jaP(ja[40]1()loglog()()jjjIaPaPa==-(3-1)I(E)EP(ja=1)I(ja)=0(3-1)211{}12,,...,NAaaa=,jaP(ja,1()1NjjPa==∑(3-2)H1()()log()NjjjjHEIaPaPa=⎡⎤==-⎣⎦∑g(3-3)H00logo=0213.2.23.2.2.13.13.1(8×8,4×4)[24]3.2.2.2m×n2212,...,,qsssq11,,...,qPPP2log(1)iP21logqiiiHPP==-∑(3-4)256256(0-255),1n×n(n=8)234(3-4)53.18×84.58×8iP(3-4)8×8675%3.18×8LenaPeperBaboonBridgeElaineAirplaneCameramanBean4.54.64.54.54.54.34.24.2233.2.2.3HVS2,1(,)kqxyBSfxyfn∈⎡⎤=-⎣⎦∑(3-5),1(,)kqxyBffxyn∈=∑kBnmax210lgfwPSNRwRMSE=(,)fxykB256×2561n×n(S)2(3-5)33.3243.3.1[25][26](Arnold)Hilbert[27]ArnoldArnold'11'12xxyy⎡⎤⎡⎤⎡⎤=⎢⎥⎢⎥⎢⎥⎣⎦⎣⎦⎣⎦(3-6)ArnoldHilbert25(1)(2)(3)[28](1)(2)(3)3