Abstract: The overlapping phenomenon often appears in the cell image recognition.It is necessary to design an effective algorithm to separate the clustering cells into single ones.Most of the existed separating algorithms are based on Mathematical Morphology，but all algorithms have limitation.An algorithm based on improved math morphology separating is presented in the paper for the problem.Using a simple and fast contour separating computation the algorithm can obtain transformed distance.Then，it uses the watershed method to separate cells.Experimental results show that the algorithm has a good separation effects.

Key words: overlapping cells, separation, distance transform, watershed algorithm

摘要： 在细胞图像识别中，经常会遇到细胞重叠的现象，需要有一种有效的方法将重叠在一起的细胞分离成单个的细胞。目前很多分离算法都是基于数学形态学的，但这些算法都存在着一定的局限性。为此，提出了一种改进的基于数学形态学的分离算法，该算法首先采用一种简单、快速的边界剥离方法得到距离变换结果，然后再用分水岭方法对细胞图像进行分离。实验结果表明，算法获得了较好的分离效果。

关键词: 重叠细胞, 分离, 距离变换, 分水岭