一种SAR相对最优极化问题研究

:2008-12-25;:2009-04-27:(1984-),,SARE-mail:[e-mail]张琼1,2,曹芳1,洪文1(1.微波成像技术国家级重点实验室,中国科学院电子学研究所,北京100190;2.中国科学院研究生院,北京100039):极化SAR运用地物目标的散射回波数据及天线收发两端的极化矢量合成功率图像目标的相对最优极化(最优极化对比增强;OPCE问题)目的在于寻找一种收发极化状态而使得期望目标和背景杂波的接收功率对比度最大化系统总结了SAR目标相对最优极化问题的解析算法与数值算法,对原有方法提出了一定程度的改进和新理解,并通过数据实验验证了算法的有效性以及合理性:相对最优极化;对比增强;OPCE;极化对比度:TN957.52+4:A:1004-0323(2009)04-0517-071引言(Po-lSAR),,[1-5](OptimizationofPolarimetricContrastEnhancement,OPCE),,,,,,Ioannidis[2]LagrangeOPCE,Cadzow[3]Swartz[4];Kong[5],MottBoerner[6],Yang[7,8],,SAROPCE,,,2OPCE2.1Lagrange,Lagrange:Wopt=[WhhWhvWvv]T(1),(2):Maxrab=WHCaWWHCbW(2)Lagrange,:f(W)=WHCaW+[1-WHCbW](3)W:f(W)/WR=0f(W)/WI=0(4):CaW=CbW(5)maxopt,WWopt:Y=WHoptX(6):X,X=[ShhShvSvv]T,Y2.2Borgeaud[10]:kB3=[Shh,2Shv,Svv]T(7)Borgeaud,:C=kB3kHB3(8)Pauli[10]:24420098REMOTESENSINGTECHNOLOGYANDAPPLICATIONVol.24No.4Aug.2009kP3=12[Shh+Svv,Shh-Svv,2Shv]T(9)Pauli,T:T=kP3k*TP3(10)TPauli,;CBorgeaud,:C=AHTA,A=10110-1020(11),,:rab=WHCaWWHCbW=WHAHTaAWWHAHTbAW=WHAHT12aT12aHAWWHAHT12bT12bHAW(12)Lagrange,:T-12bTaT-12bHe=e(13):eMoptWMopt,:I=WHMoptkP32=WHMoptTWMopt(14)Lagrange,Pauli,,Borgeaud,W=[WhhWhvWvv]T,BorgeaudkB32,,3OPCE3.1(SUMT),,Yang[8](SUMT),:min(-[K1]gh+k[K2]gh)s.t.g21+g22+g23=1h21+h22+h23=1(15),,,ghStokes,[K1][K2]Kennaugh,,3.2(OPCE),,Yang[8],,,OPCE:maxhT[K]aghT[K]bgs.t.g21+g22+g23=1h21+h22+h23=1(16),Yang:maxA0+A1g1+A2g2+A3g3B0+B1g1+B2g2+B3g3s.t.gi=Ai-rBi3i=1(Ai-rBi)2(i=1,2,3)(17):r=z12+z212-z1z2z2z1=A20-A21-A22-A23z2=B20-B21-B22-B23z12=A0B0-A1B1-A2B2-A3B3(18),gopthopt,,,,3.3(GOPCE)60,Touzi[9,10]:OPCE(GeneralizedOPCE)[9]GOPCE,:51824maxTA[3i=1xirr(TA)]2TB[3i=1xirr(TB)]2hT[K(TA)]ghT[K(TB)]gs.t.g21+g22+g23=1h21+h22+h23=1x21+x22+x23=1(19):r1r2,r3=HTATBgh,OPCExi(i=1,2,3),:TA[R(TA)]x=TB[R(TB)]x(20):[R(TA)]=[ri(TA)rj(TA)]33,[R(TB)]=[ri(TB)rj(TB)]33x=(x1,x2,x3)T,,3.4,,:(1),(2),(3),,(4),,,,,3.2,:maxCP=nP1-P2nP1+P2=nhTK1g-hTK2gnhTK1g+hTK2gs.t.g21+g22+g23=1h21+h22+h23=1(21):n0,P1P2,nCP,,P2=0P1=0,CP1;,0CP1CP,,,,3.3GOPCE4JPLAIRSAR1985SanFranciscoLSAR,1024900,554.1()()(),,121Fig.1OPCEimagesofurbanversusvegetationusingtraditionalalgorithm5194:SAR2Fig2OPCEimagesofurbanversusvegetationusingimprovedalgorithm1HHHVVV3OPCE,,,,1SanFranciscoBayTable1PolarimetriccontrastsofurbanversusvegetationinSanFranciscoBayWr(dB)HH[100]T9.18HV[010]T3.43VV[001]T7.79[1,0,-0.0222+0.0038i]T9.20[0.2952+0.1332i,0.8945,0.3078-0.0140i]T10.124.2SUMTSUMT,23,SUMT,,2SanFranciscoBaySUMTTable2CalculationresultsofurbanversusvegetationusingSUMTnumericalalgorithm(dB)g0n=[1,1,0,0]T126gm=[1,0.6303,-0.9155,0.3187]Thm=[1,-0.3984,0.9052,-0.3464]T10.55g0n=[1,-1,0,0]T126gm=[1,-0.3984,0.9052,-0.3464]Thm=[1,0.6303,-0.9155,0.3187]T10.55g0n=[1,0,1,0]T123gm=[1,-0.3984,0.9052,-0.3464]Thm=[1,0.6303,-0.9155,0.3187]T10.55g0n=[1,0,-1,0]T124gm=[1,0.6303,-0.9155,0.3187]Thm=[1,-0.3984,0.9052,-0.3464]T10.55g0n=[1,0,0,1]T109gm=[1,0.6303,-0.9155,0.3187]Thm=[1,-0.3984,0.9052,-0.3464]T10.55g0n=[1,0,0,1]T123gm=[1,-0.3984,0.9052,-0.3464]Thm=[1,0.6303,-0.9155,0.3187]T10.55520243SUMTFig.3ImagesofurbanversusvegetationusingSUMTnumericalalgorithm4.3OPCEGOPCEOPCEGOPCE,343SanFranciscoBayOPCEGOPCETable3CalculationresultsofurbanversusvegetationusingOPCEandGOPCEnumericalalgorithms(dB)OPCEGOPCEg0n=[1,1,0,0]T23gm=[1,0.7466,-0.6540,0.1223]Thm=[1,0.3525,0.9155,-0.1941]T10.2725.82g0n=[1,-1,0,0]T20gm=[1,0.3525,0.9155,-0.1941]Thm=[1,0.7456,-0.6550,0.1225]T10.2725.82g0n=[1,0,1,0]T15gm=[1,0.3514,0.9159,-0.1942]Thm=[1,0.7458,-0.6548,0.1225]T10.2725.82g0n=[1,0,-1,0]T17gm=[1,0.7457,-0.6549,0.1225]Thm=[1,0.3517,0.9158,-0.1942]T10.2725.82g0n=[1,0,0,1]T17gm=[1,0.7467,-0.6538,0.1223]Thm=[1,-0.3526,0.9154,-0.1941]T10.2725.82g0n=[1,0,0,-1]T123gm=[1,-0.3524,0.9155,-0.1941]Thm=[1,0.7465,-0.6540,0.1223]T10.2725.824OPCEGOPCEFig.4ImagesofurbanversusvegetationusingOPCEandGOPCEnumericalalgorithms,OPCE,GOPCE,GOPCExm=[-6.6555,-13.0443,7.2257]TOPCESUMT,,4.4OPCEGOPCE3.4,OPCEGOPCE,n=1,SAR5214:SAR,,,SAR5(a),,OPCEGOPCE,n=0/2n(0),nZ+,,0=0.01,OPCE0,,,,:gm=[1,0.8375,-0.0251,-0.5459]T;hm=[1,-0.8874,0.0986,0.4503]TOPCEGOPCE,13.02dB37.62dB55OPCEGOPCEFig.5ImagesofurbanversusoceanusingOPCEandGOPCEnumericalalgorithmsOPCEGOPCE,,SAR:gm=[1,0.9093,-0.0597,-0.4117]T;hm=[1,0.9082,0.0758,0.-0.4115]T,22.34dB48.33dB6,OPCEGOPCE,,,6OPCEGOPCEFig.6ImagesofurbanversusoceanusingOPCEandGOPCEimprovednumericalalgorithms522245SAR(),,,,(),Lagrange,,PauliCloude-Pottier,Pauli,SARSUMTOPCEGOPCE,,OPCEGOPCE,,:[1]KostinskiAB,BoernerWM.OnthePolarimetricContrastOptimization[J].IEEETransactiononAntennasandPropa-gation,1987,(AP-35):988-991.[2]IoannidisGA,HammersDE.OptimumAntennaPolariza-tionsforTargetDiscriminationinClutter[J].IEEETrans.AntennasPropagation,1979,(AP-27):357-363.[3]CadzowJA.Generalizeddigitalmatchedfiltering[C].South-ernSymposiumSystemTheoryVirginiaBeach,VA,1980:307-312.[4]SwartzAA,YuehHA,KongJA,etal.OptimalPolariza-tionsforAchievingMaximumContrastinRadarImages[J].JournalofGeophysicalResearch,1988,93(12):15252-15260.[5]KongJA,YuehSH,LimHH,etal.ClassificationofEarthTerrainUsingPolarimetricSyntheticApertureRadarImages[C].PolarimetricRem