一种改进的基于分水岭算法的粘连分割研究

Improved research for overlapping segmentation based on watershed algorithm

在细胞图像制作过程中,由于设备或人为原因,经常会出现细胞粘连程度分布不均匀的情况,从而影响分割的效果,给后续的统计分析造成了一定的困难。提出自适应距离变换,结合分水岭分割的处理方法,把细胞的粘连程度信息引入到距离变换函数中,根据细胞粘连程度的不同,自适应地变换距离函数的参数,从而有效地防止欠分割和过分割。实验证明,这种改进的方法能成功地分离不同粘连程度的粘连细胞,并具有良好的稳定性。

During the facture of image cell,on the reason of the equipment and something man-made,there will bring that the overlapping degree of cell distributes asymmetrically,which will infect the result of the segmentation,so it will bring difficulty to the following statistic and analysis.This article advances a new processing method,which combines watershed algorithm and selfadapt distance transform.In this method ,we add the information of the degree of overlapping to the function of distance transform. Therefore,the function of distance transform can change its parameter along with the change of the degree of the overlapping consequently,thus the over separating and the lack of separating can be avoid effectively.Experiment proves that the new method can separates the overlapping cell images successfully and steadily.