Abstract: During the facture of image cell, on the reason of the equipment and something man-made, there will be things that the overlapping degree of overlapping cell distributes asymmetrically, which will infect the result of the segmentation, so it will bring difficulty to the following statistic and analysis. This article advance a new processing method, which combine watershed algorithm and self-adapt 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 proved that the new method can separate the overlapping cell images successfully and steadily.

摘要： 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。本文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的分离不同粘连程度的粘连细胞，并具有良好的稳定性。