A separating algorithm based on granulometry for overlapping circular cell images

The overlapping phenomenon often appears in cell images. This paper describes an algorithm to estimate the shape of every single -circular cell in overlapping area. Morphological granulometry is used to estimate cell??s radius (denoted as R). Then, with the information about radius and chain code, every single cell??s central area, which is generally composed of an amount of points, is estimated. Morphological dilation with a disk-shaped structuring elemenl of radius R is applied on every single cell??s central area. The intersection of the dilation result and the original binary image is used to estimate the shape of every single cell in the overlapping area.