資源描述:
《基于方差成分?jǐn)U張壓縮的稀疏貝葉斯ISAR成像方法.pdf》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫(kù)。
1、第36卷第7期電子與信息學(xué)報(bào)V_o1_36No.72014年7月JournalofElectronics&InformationTechnologyJu1.2014基于方差成分?jǐn)U張壓縮的稀疏貝葉斯ISAR成像方法蘇伍各①王宏強(qiáng)①鄧彬①秦玉亮①凌永順②f國(guó)防科技大學(xué)空間電子信息技術(shù)研究所長(zhǎng)沙410073)(電子工程學(xué)院合肥230037)摘要:基于貝葉斯框架下的稀疏重構(gòu)方法,由于考慮了稀疏信號(hào)的先驗(yàn)信息以及測(cè)量過(guò)程中的加性噪聲,因而能夠更好地重建目標(biāo)系數(shù),然而傳統(tǒng)的稀疏貝葉斯學(xué)習(xí)fSBL)算法參數(shù)多,時(shí)效性差。該文考慮一種新
2、的稀疏貝葉斯學(xué)習(xí)方法方差成分?jǐn)U張壓縮(ExCoV1,其不同于SBL中賦予所有的信號(hào)元素各自的方差分量參數(shù),ExCoV方法僅僅賦予有重要意義的信號(hào)元素不同的方差分量,并擁有比SBL方法更少的參數(shù)。基于計(jì)算機(jī)層析成像技術(shù)框架下的ISAR成像模型,該文將ExCoV方法結(jié)合壓縮感知fCS)理論將其進(jìn)行ISAR成像,并從適用性和成像效果等方面與常用的極坐標(biāo)格式算法(PFA),卷積逆投影算法(CBPA)$1傳統(tǒng)的稀疏重構(gòu)算法進(jìn)行比較,點(diǎn)目標(biāo)仿真結(jié)果表明基于ExCoV的方法得到的ISAR像具有低旁瓣,高分辨率的特點(diǎn),真實(shí)數(shù)據(jù)的成像結(jié)果
3、表明該方法是一種比SBL更有效的ISAR成像算法。關(guān)鍵詞:逆合成孔徑雷達(dá);計(jì)算機(jī)層析成像;稀疏貝葉斯學(xué)習(xí);方差成分?jǐn)U張壓縮;稀疏恢復(fù)中圖分類號(hào):TN957.52文獻(xiàn)標(biāo)識(shí)碼:A文章編號(hào):1009.5896(2014)07—1525—07DOI:10.3724/SP.J.1146.2013.01338SparseBayesianRepresentationoftheISARImagingMethodBasedonExCoVSuWu-ge①WangHong—qiang①DengBin①Q(mào)inYu—liang①LingYong—
4、shun②①(Sch0DfofElectronicScienceandEngineering,NationalUniversityofDefenseTechnology,ChⅡngsha410073,Chin0)②(ElectronicEngineeringInstitute,Hefei230037,China)Abstract:Bytakingintoaccountofthepriorinformationofthesparsesignalandtheadditivenoiseencounteredinthemeasu
5、rementprocess,thesparserecoveralgorithmundertheBayesianframeworkcanreconstructthecoeficientbetter.However,thetraditionalSparseBayesianLearning(SBL)algorithmholdsmanyparametersanditstimelinessispoor.Inthispaper,anewsparseBayesianlearningalgorithmnamedExpansion—Com
6、pressionViance-componentbasedmethodfExC0V)iSconsidered,whichonlyendowsadifferentvariance-componenttothesignificantsignalelements.Unlikely,theSBLhasadistinctvariancecomponentontheallsignalelements.Inaddition,theExCoVhasmuchlessparametersthantheSBL.CombinedwiththeC
7、ompressSensing(CS)theory,theExCoVisusedintheISARimagingmodelundertheComputerizedTomography(CT)frame,anditsapplicabilityandtheimagingqualityarecomparedwiththePolarFormatAlgorithm(PFA),ConvolutionBackProjectionAlgorithm(CBPA)andthetraditionalsparserecoveralgorithm.
8、ThepointscattersimulationverifiesthattheInverseSAR(ISAR)imageobtainedbytheExC0Vhaslowsidelobeandhighresolution,andisnotsensitivetonoise.Theimagingresultsofreal