Automatic Segmenting Cell Nucleus with Improved Snake
Yihong Ding · Jisuanji gongcheng · 2006
This paper presents a novel cell nucleus segmentation method for esophageal cell image.Firstly the ultimate erosion operation is applied to detect the localization of each nucleus and sign it with a seed point.Then the proposed active contour model is built to track the nuclear boundaries.The proposed model describes nuclear contour with sampled points of polar coordinates.A negative constant is added as the growing energy to the energy function at the adjacent point outside when the average intensities difference between the block at this point and the block at the nuclear center point is less than a preset threshold.The growing energy can push the contour points outward.So the initial contour points need not be close to the nucleus boundary.Applying greedy algorithm,the contour points are designed to move from nuclear center to nuclear boundary along the radial directions.So the computation complexity of the model is small.