Most relevant weighted filtering with one-step singular correlation recognition.pdfVIP

Vo1．3No．4．Dec．2012 JournalofM easurementScienceand Instrumentation Sum No．12 M ostrelevantweightedfilteringwith one—step singularcorrelationrecognition LIUChao(~O] 超)，CAIJian—chao(蔡建超) (1．SchoolofMechanics，JinzhongUniversity，Jinzhong030600，China； 2．CollegeofPhysicsandElectronicsEngineering，ShanxiUniversity，Taiyuan030006，China) Abstract：Basedontherecognitionofone—step singularcorrelationandtheremedyingmethodsobtainedbefore，thecorrela． tionpropertiesoftheneighborhoodpixelsandthecharacteristicsofimagede—noisingwereanalyzed．A kindofmostrelevant weightedfilteringmethodbasedonone—stepsingularcorrelationrecognition(OSSC—MRWF)wasputforward．Thesimula— tionexperimentsweredoneandthecomparisonwithsomecommonlyusedmethodsundersalt—and—peppernoiseswasmade． Theresultsshow thattheproposedmethodcannotonlyeffectivelyrecognizesalt--and—-peppernoisesandmendup thenoise points，butalsoprotecttheoriginalinformationsuchastheedgedetailsverywel1．Theaccuracyandperformanceindicators arefurtherimprovedconsiderably． Keywords：correlation function；im agefiltering；one—step singularcorrelation；singularvaluerecognition；weightedfilter— ing；salt—and—peppernoise CLD number：TP911．73 Documentcode：A ArticleID：1674-8042(2012)04—0328—05 doi：10．3969／j．issn．1674—8042．2012．04．006 proveimagefilteringresultsthrough weightorde— 0 Introduction termining suitable sub—windows．sOmeerrorsm ight appearduring the processing and be unsuitablefor Itisunavoidableto bedisturbed bynoisesin im ． somehighqualityprocessing．One—stepsingularcor— aging，transtorming，receivingandprocessing，etc-