Design and Implementation of an Improved Overlapped Cell Image Segmentation Method

This paper presents an improved segmentation method for overlapped cell image. The method first implements multi-scale edge detection to the cell image, and distance transform is applied to get the pixel location, then the edge detection result and distance transform result are fused for watershed segmentation, and finally region merging is implemented to remove over-segmentation in the image. Experimental results show that the method can separate conglutination cell effectively, and get accurate and closed region contours.