Cells images color segmentation based on thresholding and watershed segmentation

Abstract

In this work, we propose a methodology of biomedical image segmentation. We consider two stages of segmentations, every one uses a chromatic plan appropriated to the RGB space, the goal of the first segmentation is to isolate all tumorous zone. We have tested three bi-level thresholding methods: Otsu method, convex hull method and fuzzy method. The second segmentation has for goal the insulation of tumorous cell of the tumorous zone. We use watershed method for this purpose which is an adapted method for biomedical image. We can appreciate the successful segmentation at 90%.