欢迎来到天天文库
浏览记录
ID:55398701
大小:1.02 MB
页数:5页
时间:2020-05-15
《基于NSCT和全变差模型的医学图像去噪.pdf》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、第32卷第5期应用科学学报V01.32No.52014年9月JOURNALOFAPPLIEDSCIENCES——ElectronicsandInformationEngineeringSept.2014DOI:10.3969/j.issn.0255—8297.2014.05.008MedicalImageDenoisingUsingNon—subsampledContourletTransformandT0talV_ariationModelMAXiu—li,-.ZHOUFeng,-.ZHOUXiao-jun,1.SchoolofCommunicationandInformat
2、ionEngineering,ShanghaiUniversity,Shanghai20O4,China2.InstituteofSmartCity,ShanghaiUniversity,Shanghai20O4,ChinaAbstraetThecharacteristicsofnon—subsampledContourlettransform(NSCT)andtotalvariation(TV)modelingareanalyzed.AmixedmodelofNSCTandTVisappliedtomedicalimagedenoisinginthispaper.NSCTfi
3、lter—baseddecompositionofnoisymedicalimagesisperformed.AninitialdenoisedimageisproducedusingaVisushrinkthresholdalgorithm.ThefinaldenoisedimageisobtainedbyprocessingtheinitialdenoisedimagewiththeTVmode1.Experimentalresultsshowthattheimagedetailsarewellpreservedbyusingtheproposedmethod.Bothpe
4、aksignal—to-noiseratio(PSNR)andvisualqualityaresuperiortosomeotherdenoisingalgorithms.Keywords:non—subsampledContourlettransform,totalvariation,medicalimagedenoising,peaksignal-to-noiseratio(PSNR)基于NSCT和全变差模型的医学图像去噪马秀丽,周峰,一,周小军,21.上海大学通信与信息工程学院,上海2004442.上海大学智慧城市研究院,上海200444摘要:分析了非下采样Contour
5、let变换(nonsubsampledContourlettransform,NSCT)和全变差模型的特点,提出将NSCT和全变差混合模型应用于医学图像去噪.首先,通过NSCT变换将含噪图像分解,运用Visu萎缩阈值将NSCT系数进行处理,得到初次去噪图像.然后,采用全变差模型对初次去噪图像进一步处理得到最终去噪图像.实验结果表明:该方法可以很好地保留图像细节,无论在客观上的峰值信噪比还是主观上的视觉效果都优于其他去噪方法.关键词:非下采样Contourlet变换;全变差;医学图像去噪;峰值信噪比中图分类号:TP391文章编号:02558297(2014)05—0481—05I
6、nmedicalimageacquisitionandtransmissionisticsoftheuniquedata.Ithaspoordirectionselec.process.noi‘seisinevitablyintroducedintomedicaltivityanditcan’tcaptureedgeinformationoftheim.images.Thecontaminatedimagesinfluencethediagagesefectively.Indenoisingprocess.itisinevitablenosis.soit’snecessaryt
7、oremovethenoiseofmedicaltointroduceacertaindegreeoffuzzyatedgesandde—images.Mostofthecurrentimagedenoisingmethodstailtextures.Tosolvethisproblem,thenonsubsam—arebasedonwavelettransform[1J.Butwavelettrans—pledContourlettransformfNSCT)wasproposedasaf
此文档下载收益归作者所有