一种新的基于Snake模型的方法及病理细胞图像分割应用

A NEW METHOD BASED ON SNAKE MODEL WITH APPLICATION ON PATHOLOGICAL CELL IMAGE SEGMENTATION

Snake模型在图像分割中对于初始轮廓的设置要求高,并且难以逼近凹陷区域。设计了动态外部力,提出了奇偶点对规则,对于轮廓线上的任意关键点都能够计算出其所受到的动态外部力方向,简洁而有效地解决了这两个问题。此外,讨论并确定了轮廓线停止运动的条件,以及无效轮廓线的去除方法。用于病理细胞图像分割测试结果表明效果很好。

When used in image segmentation, Snake model demands the location of the initial contour to be near the object boundary and it is hard to simulate the concavity. The Dynamic External Force （DEF） is designed,in which Odd-Even Twin Rule is used to calculate the direction of the DEF of any point along the contour so as to solve the two problems effectively and simply. Moreover,the stop criteria of the contour＇s movement as well as the removing of invalid contours is discussed and defined. The revised Snake Model has proved to get very good results on pathological cell image segmentation.