中文摘要：

为使细胞分割的结果更加精确,提出一种基于伪中值双边滤波和水平集函数的细胞图像分割方法.首先使用伪中值双边滤波对图像进行预处理,然后利用水平集方法对改进的CV模型进行两次曲线演化,分别得到细胞质与背景分界线,细胞核与细胞质分界线.结果表明：伪中值双边滤波在减弱高斯噪声的同时,同时去除了椒盐噪声,但没有弱化边界,LCV模型在CV模型的基础上添加了局部项,使得对于灰度不均匀的图像分割效果较好.结论：在使用水平集方法进行图像分割之前先进行伪中值双边滤波,同时为CV模型添加局部项,能够增强细胞分割结果的准确性.

英文摘要：

In order to obtain more accurate data on cell division, we proposed an image segmentation method based on level set function and pseudomedian bilateral filtering. Cell image was pre-processed by pseudomedian bilateral filtering, then the improved CV model was evoluted twice with level set method. Contour between background and cytoplasm, contour between cytoplasm and nucleus could all be obtained. It has been found that pseudomedian bilateral filtering removed gauss and salt-and-pepper noises, without weakening marginal information. LCV model was added local item on traditional CV model to obtain more accurate segmentation results in cell image with uneven gray levels. This algorithm improved significantly accuracy of cell segmentation. It is concluded that pseudomedian bilateral filtering and addition of local item to CV model could enhance the accuracy of cell image segmentation.