非线性自适应迭代重建算法用于相干层析抗噪仿真.pdf

《非线性自适应迭代重建算法用于相干层析抗噪仿真.pdf》由会员上传分享，免费在线阅读，更多相关内容在应用文档-天天文库。

1、光电子·澈光第24卷第8期2013年8月JournalofOptoelectronics·LaserVo1．24No．8August2013非线性自适应迭代重建算法用于相干层析抗噪仿真曲培树，宋一中(德州学院物理系，山东德州253000)摘要：为进一步完善非线性自适应迭代重建算法(NAIRT)并将其可靠地应用于实际流场层析诊断，仿真研究了NAIRT的抗噪性能。采用仿真技术，模拟含跳变的复杂空气流场，相干投影，获得Radon变换投影数据。采用RandG函数模拟产生Gauss随机噪声，噪声的相对强度通过信噪比(sNR)标征，设置强度不同的SNR，获得相应的随机噪声。随机噪声与投影数据线性叠加，获得

2、含噪投影。采用NAIRT逆投影重建，重建结果与无噪重建结果比较，并用均方差(MsE)对重建结果定量分析。结果发现，在60dBSNR噪声情况下，NAIRT能够比较精确重建仿真流场，MsE降到1．82×10；10dBSNR时，NAIRT仍能重建出仿真场的轮廓，MSE稳定在4．91×1O；100dBSNR时，NAIRT能够十分精确地重建仿真流场，MSE衰减到1．26×106，逼近无噪重建效果。重建图像和误差分析都表明，NAIRT具有十分优越的抗噪性能。关键词：非线性自适应迭代重建算法(NAIRT)；抗噪性能；随机噪声；信噪比(SNR)；投影；重建中图分类号：0438文献标识码：A文章编号：1005—

3、0086(2013)08—1644—07SimulationOilanti-noiseperformanceofnonlinearauto-adjustingiterativereconstructiontechniqueusedininterferometrictomo。·graphyQuPei-shu，SONGYi-zhong(DepartmentofPhysics，DezhouUniversity，Dezhou253000，China)Abstract：Inordertofurtherimprovenonlinearauto-adjustingiterativereconstructio

4、ntechnique(NAIRT)andreliablyapplyittodiagnoseactualairflowfield，theanti-noiseperformanceoftheNAIRTisspeciallyinvestigatedbysimulations．Acomplicatedairflowfieldissimulatedwithtransitionobject．Itisprojectedbyinterferometrictomography．TherealprojectionsareobtainedbyRadontransformation．Gaussianrandomnoi

5、sesignalsaresimulatedtobeproducedwithfunctionRandG．ThenoiseSrelativeintensityisexpressedwiththeratioofsignaltonoise(s／N)．AseriesofcorrespondingGaussianrandomnoisesignalsareproducedwithdifierentintensitylevelsbysimplecalculation．Theprojectionscontainingnoisesareobtainedfromtherealprojectionsaddedwith

6、noise1inearly．Theinverseprojectionreconstruc—tionsareimplementedwithNAIRT．Thereconstructedresultsareanalyzedinmeansquareerror(MSE)indexandcomparedwiththeresultreconstructedwithcleanprojectionsunderthesameconditions．Atthenoiselevelof6OdBS／N，NAIRTcouldreconstructthemodelbyadecentaccuracy．TheMSEdec—lin

7、stO1．82×10—5afterabout40iterationcycles．Atthenoiseleve1of10dBS／N，NAIRTcouldstil1reconstructthemodelbytheleveloftheprofile．TheMSEiSstabilizedat4．91×10afterabout40it—erationcycles．Atthenoise1eve1of100dB