摘要
Anewwavelet-basedimagedenoisingalgorithm,whichexploitstheedgeinformationhiddeninthecorruptedimage,ispresented.Firstly,acanny-likeedgedetectoridentifiestheedgesineachsubband.Secondly,multiplyingthewaveletcoefficientsinneighboringscalesisimplementedtosuppressthenoisewhilemagnifyingtheedgeinformation,andtheresultisutilizedtoexcludethefakeedges.Theisolatededgepixelisalsoidentifiedasnoise.Unlikethethresholdingmethod,afterthatweuselocalwindowfilterinthewaveletdomaintoremovenoiseinwhichthevarianceestimationiselaboratedtoutilizetheedgeinformation.Thismethodisadaptivetolocalimagedetails,andcanachievebetterperformancethanthemethodsofstateoftheart.
出版日期
2006年02月12日(中国期刊网平台首次上网日期,不代表论文的发表时间)