Automatic segmentation for cell images based on bottleneck detection and ellipse fitting

To segment the overlapping cells in microscopic images, an automatic method for cell image segmentation based on bottleneck detection and ellipse fitting is proposed. Firstly, cell image is segmented by threshold method, followed by a polygonal approximation to extract the feature points of cell edge. Secondly, candidate splitting point pairs are obtained by calculating the bottleneck rate between each feature point pair, and further judged by ellipse fitting to find the correct splitting point pair. Then, a cell is separated from the overlapping cells according to the splitting point pair, and the remaining edge is patched up to form a new closed contour by an improved ellipse fitting method. Finally, repeat the above steps on the new closed contour until all cells are separated. The performance of this method is evaluated on the blood and fluorescent cell databases. Experimental results show that the proposed method can effectively segment overlapping cells with high accuracy and less time, which is superior to many existing methods.