No abstract

No abstract

No abstract

No abstract

No abstract

No abstract

No abstract

The graph cuts in image segmentation have been widely used in recent years because it regards the problem of image partitioning as a graph partitioning issue, a well-known problem in graph theory. The normalized cut approach uses spectral graph properties of the image representative graph to bipartite it into two or more balanced subgraphs, achieving in some cases good results when applying this approach to image segmentation. In this work, we discuss the normalized cut approach and propose a Quadtree based similarity graph as the input graph in order to segment images. This representation allow us to reduce the cardinality of the similarity graph. Comparisons to the results obtained by other graph similarity representation were also done in sampled images.