简介:Withtherapiddevelopmentofdeepsubmicron(DSM)VLSIcircuitdesigns,manyissuessuchastimeclosureandpowerconsumptionaremakingthephysicaldesignsmoreandmorechallenging.InthisreviewpaperweprovidereaderswithsomerecentprogressoftheVLSIphysicaldesigns.Therecentdevelopmentsoffloorplanningandplacement,interconnecteffects,modelinganddelay,bufferinsertionandwiresizing,circuitorderreduction,powergridanalysis,parasiticextraction,andclocksignaldistributionarebrieflyreviewed.
简介:Inthefieldofmedicalimaging,thetraditionallocalbinarypattern(LBP)anditsimprovedalgorithmsareoftensensitivetonoise.TraditionalLBPsaresolelybasedonthesignalinformationfromlocaldifferences,andthebinaryquantizationmethodoversimplifiesthelocaltexturefeatureswhiledisregardingtheimaginginformationfromtheconcaveconvexregionsbetweenthehigh-orderpixelsandtheneighboringsamplingpoints.Therefore,weproposeanimprovedDerivedMeanCompleteLocalBinaryPattern(DM_CLBP)algorithmbasedonhigh-orderderivatives.IntheDM_CLBPmethod,thegreyvalueofasinglepixelisreplacedbythemeangreyvalueoftherectangularareablock,andthedifferencebetweenpixelvaluesintheareaisobtainedusingthesecond-orderdifferentiationmethod.Basedonthecalculationconceptofthecompletelocalbinarypattern(CLBP)algorithm,thecascadesignsandmagnitudesofthetwocomponentsareencodedandrecombinedinDM_CLBPusingauniformpattern.TheresultsfromtheexperimentsshowedthattheproposedDM_CLBPdescriptorsachievedaclassificationaccuracyof94.4%.ComparedwithLBPandotherimprovedalgorithms,theDM_CLBPalgorithmpresentedinthisstudycaneffectivelydifferentiatebetweenlesionareasandnormalareasinthyroidMR,(magneticresonance)imagesandshowstheimprovedaccuracyofareaclassification.