REGION GROWING SEGMENTATION OF TEXTURED CELL IMAGES

The authors introduce a region growing algorithm working in conjunctio n with the K-means algorithm for the segmentation of textured cell ima ges. The iterative algorithm initially segments the most easily classi fiable regions which are grown in subsequent iterations. Only the most easily classifiable pixels in the remaining regions are segmented in each iteration. Experimental results of the procedure in segmenting br east cancer cells are demonstrated and compared to the K-means algorit hm.