Active contours driven by local Gaussian distribution fitting energy

代码说明：

This paperpresentsanewregion-basedactivecontourmodelinavariationallevelset formulationforimagesegmentation.Inourmodel,thelocalimageintensitiesare describedbyGaussiandistributionswithdifferentmeansandvariances.Wedefinea local Gaussiandistributionfittingenergywithalevelsetfunctionandlocalmeansand variancesasvariables.Theenergyminimizationisachievedbyaninterleavedlevelset evolution andestimationoflocalintensitymeansandvariancesinaniterativeprocess. The meansandvariancesoflocalintensitiesareconsideredasspatiallyvarying functionstohandleintensityinhomogeneitiesandnoiseofspatiallyvaryingstrength (e.g. multiplicativenoise).Inaddition,ourmodelisabletodistinguishregionswith similarintensitymeansbutdifferentvariances.Thisisdemonstratedbyapplyingour methodonnoisyandtextureimagesinwhichthetexturepatternsofdifferentregions can bedistinguishedfromthelocalintensityvariance.Comparativeexperimentsshow the advantagesofth