An improved fuzzy c-means clustering algorithm with guided filter for Image Segmentation

Abstract

Fuzzy c-means clustering with guided filter (FCM+GF) is an effective method for noisy image segmentation. However, the parameter ε of guided filter in the FCM+GF is set to a fixed value, which weakens the ability of the FCM+GF to partition images with different noise rates. In this paper, an improved fuzzy c-means with guided filter method (FCM+GF_I) is proposed. In our method, a new influence factor ρ is defined to adjust the guidance image. By adjusting the value of ρ, the proposed FCM+GF_I method achieves good performance on different noisy images. Experiments on Brain MR images show the superiority and efficiency of our method.