USINGWATERSHED TRANSFORM AND LEVELSETMETHOD
Miaomiao Liu · 2007
A novel method forimagesegmentation isproposed inthis paper, whichcombines thewatershed transform and region-based level setmethod. Thewatershed transform is first usedtopresegment theimagesoastogettheinitial partition ofit.Someuseful information oftheprimitive regions andboundaries canbeobtained. Theregion-based levelsetmethodisthenapplied forextracting the boundaries ofobjects onthebasis ofthepresegmentation. Theconsumed timedoesnotdepend onthesizeofthe imagebutthenumberofpresegmented regions because onlylabel level setfunction isupdated instead ofthelevel setfunction foreachpixel. Therefore, theproposed method iscomputationally efficient. Moreover, thealgorithm can localize theboundary oftheregions exactly duetothe edgesobtained bythewatersheds. Theefficiency and accuracy of thealgorithm isdemonstrated by the experiments ontheMR brain images. IndexTerms- imagesegmentation, watershed transform, level setmethod, region, MR brain image