为对显微粘连细胞图像进行分割,该文提出1种基于数学形态学运算的迭代腐蚀方法。对二值化细胞图像进行迭代腐蚀得到细胞种子点图像,通过分水岭分割提取细胞分割边界,形成最终的分割图像。在每一次腐蚀后得到的图像中保存细胞种子点,在整个迭代腐蚀过程中避免细胞种子点被误删,为后续使用分水岭分割进行图像分割提供了可靠的先验信息。实验表明,该算法能较好地解决距离变换方法中的过分割问题并改善极限腐蚀方法中的欠分割问题。
To segment microscopic adherent cell images ,a new iterative erosion method is presented based on mathematical morphology operation .Iterative corrosion is performed on a binary cell image to generate a cell seed point image .Cell segmentation boundaries are extracted by watershed segmentation ,and a final segmented image is obtained .Cell seed points are kept in every corrosion image to avoid deleting the cell seed points in error ,and reliable prior information is provided for the subsequent marked watershed segmentation .Experiments show that the algorithm can relieve the over-segmentation problem of the distance transform method and the under-segmentation problem of the limit corrosion method .