The segmentation of cervical cells is the core part of the automatic analysis system of cervical smear. In this paper, we propose a method for segmentation of EDF images of overlapping cervical cells based on similarity measure. Based on the consistency of focus and position information between cytoplasm and target nucleus, the similarity measure of cytoplasm is constructed by using the characteristic of EDF image which is composed of several images with different focal points. Based on similarity measure, the cytoplasmic contour was tracked to achieve rough segmentation. Finally, regularized level set evolution (DRLSE) was used to refine the obtained contour at the pixel level. The approach was evaluated based on the public dataset-ISBI2015 which is superior to existing methods in image segmentation of EDF for overlapping cells.