TheK–puterScienceDepartmentTheTechnion–IsraelInstituteoftechnologyHaifa32000,IsraelSPIE–WaveletsXISan-Diego:August2nd,2005*JointworkwithMichalAharonFreddyBruckstein*1DesignofDictionariesforRedundantandSparserepresentationofSignalsNOTE:Thislectureconcentratesonthefundamentalsofthiswork,&Sparsity2. TheQuestforaDictionary–monApproaches?3. TheQuestforaDictionary–Practice IntroducingtheK-SVD4. Results etoSparseland2DesignofDictionariesforRedundantandSparserepresentationofSignalsGeneratingSignalsinSparselandMKNAfixedDictionaryEverycolumninD(dictionary)isaprototypesignal(Atom).Thevectorisgeneratedrandomlywithfewnon-&randomvectorN3DesignofDictionariesforRedundantandSparserepresentationofSignalsSimple:binationoffewatomsfromourdictionaryDRich:Ageneralmodel:theobtainedsignalsareaspecialtypemixture-of-Gaussians(orLaplacians).Popular:essfullydeploysittoapplications. Sparselandishere!?SparselandSignalsAreInterestingMultiplybyDM4DesignofDictionariesforRedundantandSparserepresentationofSignals?OURFOCUSTODAY!!Practicalwaystoget?SignalProcessinginSparseland4MajorQuestionsIsorevenclose?Howeffective?HowdowegetD?RecentresultsgiveoptimisticanswerstothesequestionsMultiplybyDAsparse&randomvectornoiseM5DesignofDictionariesforRedundantandSparserepresentationofSignalsAgendaAVisittoSparselandMotivatingredundancy&Sparsity2. TheQuestforaDictionary–monapproaches?3. TheQuestforaDictionary–Practice IntroducingtheK-SVD4. Results Preliminaryresultsandapplications6DesignofDictionariesforRedundantandSparserepresentationofSignalsProblemSettingMultiplybyDMGivenasetofdataexamples,weassumetheSparselandmodelandask7DesignofDictionariesforRedundantandSparserepresentationofSignalsChooseD–ModelingApproachReplacethemodelwithanother,well-definedsimplemathematical,model(-wiseC2smoothregionswithC2smoothedges)ordingly,: Curvelet[Candes&Donoho] Contourlet[Do&Vetter
KSVD的英文ppt 来自淘豆网www.taodocs.com转载请标明出处.