当前位置:首页 > 商业/管理/HR > 管理学资料 > SAR场景图像中车辆目标的自动识别
317Vol.31No.720097SystemsEngineeringandElectronicsJul.2009:10012506X(2009)0721596206:2008203218;:2008209203:(19802),,,SARE2mail:panzhuo_iecas@163.comSAR1,2,1,1,3,1,2(1.,100190;2.,100039;3.,100875):(syntheticapertureradar,SAR),,,SAR,,,MSTARSAR,,,:;;;;:TP753:AAutomaticvehicletargetrecognitioninfullSARimagescenesPANZhuo1,2,GAOXin1,WANGYan2fei1,WANGBin2hui3,XIEJian2hong1,2(1.Inst.ofElectronics,ChineseAcademyofSciences,Beijing100190,China;2.GraduateSchooloftheChineseAcademyofSciences,Beijing100039,China;3.SchoolofMathematicalSciences,BeijingNormalUniv.,Beijing100875,China)Abstract:Anewmethodispresentedforautomatictargetrecognitioninfullsyntheticapertureradar(SAR)imagescenes.AsSARtargetclassificationissensitivetotargetsazimuth,basedoncorrelationfiltertheoryandkernelfeatureanalysis,anonlinearcorrelationfilterisprovided,itcantolerateadistortionoftargetsazimuth.Thenovelfilterexploitseigenvectorstoreducethedependenceofthetrainingsetandextendsthelin2earcombinationofeigenvectorsnonlinearlytoimprovetheclassificationperformance.Moreover,tokeepthecomputationtractableinhighdimensionalspace,thekernelfunctionisemployed.ThetestsusingMSTARda2tabasedemonstratetheschemeispracticalandthenovelfilterimplementstargetclassificationefficientlyandac2curatelywithouttemplatesandtargetposesestimation.Keywords:syntheticapertureradarimage;targetdetection;automatictargetrecognition;correlationfil2ter;kernelfunction0(automatictargetrecognition,ATR)SAR,SARATR[123](regionofinterest,ROI),ROI,,SARATR1,,,ATR,,,SAR,,SAR[3],,[4],,SAR[528],SAR,,,SAR7:SAR1597,,,SARATR1SARATR1,,SAR,SAR,[1](constantfalsealarmrate,CFAR),()SAR,CFARXt-^c^cTh,XtXt-^c^cTh,Xt(1),Xt;^c^c;Th,CFAR,^c^c,,2,ROI,100%,,,,,ROI,,,Px,PsumPx,Pth,PsumPth,PxPsumPth,Px(2)SAR,,,,,ROIS,[S1,S2],S,S1SS2S,(3),,ROI,SAR3ROI,,[9](syntheticdiscriminantfunction,SDF),SDF,,(1,),22D2DFT,,,2159831,N,dd2D2DFT[6](lexico2graphically)d2,,d2NX,u1N,hd2MACE[10](minimumaveragecorrelationenergy)MI2NACE[11](minimumnoiseandaveragecorrelationenergy)SDF,XHh=u(4)1,HMACE,,MINACE,,,Lagrange2D2DFT(),,,EMINACE[5](eigen2MINACE)(4),,,[5]hEMINACE=T-1/2Nei=11NiiHiT-1/2Xu(5),T;iiXp(Xp=T-1/2X)C=XpXTp;Ne(NeN)(5)EMINACEhEMINACE=Nei=1iT-1/2i(6),i=piNi(1+Ni);pi=Hiu,iV=XHpXp(6),EMINACE(T-1/2i),,,SAR,2D2DFT[15](),SAR,,,,,EMINACE,KEMINACE(kernel2EMINACE),:,X={x1,x2,,xN}(xiRd2,N)Rd2FRd2F(7),F(xi),,F,,,2D2DFT,,F,Mercer[12]K(x,y)K(x,y)=((x)(y))(8),,,Mercer[13],()K(a,b)=(a,b+1)p(9),pK(a,b)=exp(-a-b2/22)(10),2,([0,1]),,(5)hEMINACE=T-1/2Nei=11NiiHiT-1/2Xu=1NT-1/2Nei=1iiHiiT-1/2Xu=1NT-1/2HT-1/2Xu(11)=11,22,,NeNeTYRd2(11),EMINACEYhEMINACE=1NYHT-1/2HT-1/2Xu=7:SAR15991N(HT-1/2Y)(HT-1/2X)u(12),F,,KEMINACEYhKEMINACE=(Y)(hEMINACE)=1N((H)T-1/2Y)(HT-1/2X))u=1NK(Yep,Xep)u(13),Yep=HT-1/2Y,Xep=HT-1/2X,MACE,MINACEEMINACE,KEMINACE,1,(13)YhKEMINACE=1NK(Yep,Xep)K(Xep,Xep)-1u(14),K(Xep,Xep)-1()KEMINACE(NeN),,,,,,,4PIV3.0GHzCPU,2GB,Win2dowsServer2003,VisualC++6.04.1KEMINACE,MACE,MINACEEMINACEDARPA/AFRLSARMSTAR(movingandstationarytargetacquisitionandrecognition)[14],XHH9.6GHz,0.3m0.3mSAR,128128,SAR17[628],15,KEMINACE,(rangeprofile),3,SARa=D/2,D,=/D,,a=/2,0.3m,9.6GHz,=3.110-2/(20.3)=0.052radD2.986(0,6,,360),6(3,9,,357),,,SAR173(7)3,1,6013123T72sn_132T72sn_812T72sn_S7BTR70sn_c71BTR70sn_c71BTR70sn_c71BMP2sn_c21BMP2sn_9563BMP2sn_95666464,2D2DFT,4096,60X(409660),60Y(409660)EMINACEKEMINACENe=50,KEMINACEp=233,3,,,,,,425469410132MACET72BTR70BMP2T7255.0%15.0%30.0%BTR7018.3%68.3%13.4%BMP226.7%18.3%55.0%3MINACET72BTR70BMP2T7266.7%11.7%21.6%BTR708.3%76.7%15.0%BMP215.0%6.7%78.3%4EMINACET72BTR70BMP2T7266.7%11.7%21.6%BTR708.3%76.7%15.0%BMP216.7%6.6%76.7%5KEMINACET72BTR70BMP2T72100%0%0%BTR700%98.3%1.7%BMP20%0%100%1600316MACET72BTR70BMP2T7255.0%21.7%23.3%BTR7011.7%61.7%26.6%BMP213.3%15.0%71.7%7MINACET72BTR70BMP2T7258.3%20.0%21.7%BTR701.7%71.7%26.7%BMP25.0%6.7%88.3%8EMINACET72BTR70BMP2T7258.3%20.0%21.7%BTR701.7%71.7%26.7%BMP25.0%6.7%88.3%9KEMINACET72BTR70BMP2T7295.0%0%5.0%BTR701.7%98.3%0%BMP20%0%100%10MACET72BTR70BMP2T7253.3%15.0%31.7%BTR7017.7%65.0%18.3%BMP213.3%23.3%63.3%11MINACET72BTR70BMP2T7268.3%10.0%21.7%BTR7015%70.0%15.0%BMP210.0%8.3%81.7%12EMINACET72BTR70BMP2T7265.0%6.7%28.3%BTR701.7%85.0%13.3%BMP21.7%10.0%88.3%13KEMINACET72BTR70BMP2T7288.3%1.7%10.0%BTR700%100%0%BMP20%0%100%,,T72,BMP2BTR70,MACE,53%72%,MINACEEMINACE,58%88%KEMINACE,88%100%4.2SARSARATR,317481478,,14,9,7,53513SAR143/()1T72sn_13252T72sn_812463T72sn_S7884BTR70sn_c711485BTR70sn_c711836BTR70sn_c712257BMP2sn_95662738BMP2sn_95633259BMPsn_c21351,SARRayleigh,10-2,CFAR4,100%,4CFAR7:SAR16014,1111,20,[250,400]ROI,,36464,ROI,55ROI5,ROI(5(b)(c)(d)(f)(g)(h)),ROIKEMINACE,ROI,,,6,,15615ROI()1(552,749)349T72sn_1322(566,9)306Clutter3(604,318)416BMP2sn_95664(611,1328)395Clutter5(639,1117)405T72sn_8126(655,66)413Clutter7(790,252)256Clutter8(800,185)323Clutter9(1172,303)471T72sn_S710(1212,858)302BTR70sn_c7111(1260,1248)378BTR70sn_c7112(1447,1253)266BTR70sn_c7113(1624,405)385BMP2sn_956614(1660,1024)315BMP2sn_956315(1674,1284)391BMPsn_c2115,KEMINACE,65,BMP2,6105SAR,,,,,0360,88%100%,,,;,;SAR;SAR,(extendedoperatingcondit
本文标题:SAR场景图像中车辆目标的自动识别
链接地址:https://www.777doc.com/doc-631548 .html