简介:Thispaperproposesanovelexemplar-basedmethodforreducingnoiseincomputedtomography(CT)images.Intheproposedmethod,denoisingisperformedoneachblockwiththehelpofagivendatabaseofstandardimageblocks.Foreachnoisyblock,itsdenoisedversionisthebestsparsepositivelinearcombinationoftheblocksinthedatabase.Weformulatetheproblemasaconstrainedoptimizationproblemsuchthatthesolutionisthedenoisedblock.Experimentalresultsdemonstratethegoodperformanceoftheproposedmethodovercurrentstate-of-the-artdenoisingmethods,intermsofbothobjectiveandsubjectiveevaluations.
简介:分析对比了三种光交叉连接结点的架构.并对当前主要的波带交换技术进行了讨论.我们同时对多粒度光交换机在下一代网络中的应用进行了探讨.