Method for cell image segmentation based on bilateral filtering and CV Model

摘要

Objectives: Change the initial value of CV model to achieve the purpose of cell image segmentation fast and accurately. Material and methods: This paper selects slice image of cervical-cancer cells under a microscope as experimental materials. First, the original image is bilateral filtered and then the image is preprocessed using Otsu method to get the rough contour of cytoplasm. Then use Otsu method twice on the cytoplasm to get the rough contour of nucleus. Finally, regard the preprocessed results as the initial value of CV model and evolve the curve with level set method to obtain the final contour. Results: This proposed method costs 17.057s and the iterations are 50, the contour is accurate. Meanwhile, the existed method costs 45.329s and the iterations are 90 and it doesn't iterate to the accurate result. Conclusions: It can obtain accurate cell contour fast regarding the preprocessed result of bilateral filtering and Otsu method as the initial value of CV model.