Semi-automatic watershed merging method

Abstract

Watershed transformation frequently produces over-segmented images. The authorspropose a solution to this problem. It utilizes hierarchical cluster analysis for grouping watershedswhich are treated as objects characterized by a number of attributes. Initially the watershed mergingmethod was meant only for gray-scale images, but later it was adapted for colour images. This paperpresents further extension of the method that allows it to either automatically select the numberof classes or to provide a hint as to which numbers in a specified range should be considered first.Segmentation quality assessment functions for colour images are presented. The results obtained usingan extended watershed merging method are discussed. The examples of segmentations selected by themethod, along with the graphs of assessment functions, are shown.