A general framework for image segmentation
Rui Pires, Patrick DE SMET, Wilfried R. Philips · 2003
In this paper we describe a methodology to generate a partition of an image and how a hierarchical region merging scheme can be used to improve the quality of the segmentation. The segmentation method is based on the watershed transform. Prior to the actual segmentation, the image is smoothed to decrease the amount of detail detected by the watershed transform. To further improve the segmentation result, we use an iterative region merging process that sequentially merges the most similar pair of regions according to a pre-defined similarity metric. We propose the use of a combined similarity function that considers not only the intensity similarity of two neighboring regions but also the gradient magnitude at their dividing boundary. Results obtained illustrate the overall good performance of this segmentation methodology and the usefulness of the combined similarity function.