Segmentation via morphological granulometric features is based on fitting structuring elements into image topography from below and above. Each structuring element captures a specific tex- ture content. This paper applies granulometric segmentation to digi- tized mmmammograms in an unsupervised framework. Granulometries based on a number of flat and nonflat structuring elements are com- puted, local size distributions are tabulated at each pixel, granulometric-moment features are derived from these size distribu- tions to produce a feature vector at each pixel, the Karhumen-Loeve transform is applied for feature reduction, and Voronoi-based clus- tering is performed on the reduced Karhunen-Loeve feature set.
展开▼