简介:Inthispaper,themorphologicalfilterbasedonparametricedgedetectionispresentedandappliedtoimagingladarimagewithspecklenoise.ThisalgorithmandLaplacianofGaussian(LOG)operatorarecomparedonedgedetection.Theexperimentalresultsindicatethesuperiorperformanceofthiskindoftheedgedetection.
简介:Toachievelowerassemblydelayatopticalburstswitchingedgenode,thispaperproposesanapproachcalledcurrentweightlengthprediction(CWLP)toimproveexistingestimatemechanisminburstassembly.isintroducedtomakeadynamictradeoffbetweenthecurrentandpasttrafficunderdifferentoffsettime.SimulationresultsshowthatCWLPcanachieveasignificantimprovementintermsoftrafficestimationinvariousoffsettimeandofferedload.
简介:Anewwavelet-basedimagedenoisingalgorithm,whichexploitstheedgeinformationhiddeninthecorruptedimage,ispresented.Firstly,acanny-likeedgedetectoridentifiestheedgesineachsubband.Secondly,multiplyingthewaveletcoefficientsinneighboringscalesisimplementedtosuppressthenoisewhilemagnifyingtheedgeinformation,andtheresultisutilizedtoexcludethefakeedges.Theisolatededgepixelisalsoidentifiedasnoise.Unlikethethresholdingmethod,afterthatweuselocalwindowfilterinthewaveletdomaintoremovenoiseinwhichthevarianceestimationiselaboratedtoutilizetheedgeinformation.Thismethodisadaptivetolocalimagedetails,andcanachievebetterperformancethanthemethodsofstateoftheart.