HYBRID IMAGE SEGMENTATION USING WATERSHEDS AND FAST REGION MERGING

A hybrid multidimensional image segmentation algorithm is proposed, wh ich combines edge and region-based techniques through the morphologica l algorithm of watersheds. An edge-preserving statistical noise reduct ion approach is used as a preprocessing stage in order to compute an a ccurate estimate of the image gradient. Then, an initial partitioning of the image into primitive regions is produced by applying the waters hed transform on the image gradient magnitude. This initial segmentati on is the input to a computationally efficient hierarchical (bottom-up ) region merging process that produces the final segmentation. The lat ter process uses the region adjacency graph (RAG) representation of th e image regions. At each step, the most similar pair of regions is det ermined (minimum cost RAG edge), the regions are merged and the RAG is updated. Traditionally, the above is implemented by storing all RAG e dges in a priority queue. We propose a significantly faster algorithm, which additionally maintains the so-called nearest neighbor graph, du e to which the priority queue size and processing time are drastically reduced. The final segmentation provides, due to the RAG, one-pixel w ide, closed, and accurately localized contours/surfaces. Experimental results obtained with two-dimensional/three-dimensional (2-D/3-D) magn etic resonance images are presented.