Image segmentation by multi-level thresholding based on C-means clustering algorithms and fuzzy entropy

C-means clustering algorithms (hard C-means clustering algorithm and fuzzy C-means clustering algorithm) are common thresholding segmentation techniques. They can be easily applied to image segmentation tasks with multi-feature and multi-threshold, but the results are very sensitive to noise. In this paper, an image segmentation method combining the C-means clustering algorithms and fuzzy entropy is presented. Firstly, pre-segmentation by multi-level thresholding was made by one of the C-means clustering algorithms; then further processing was done using fuzzy entropy. Experimental results show that the proposed method can overcome the weakness of the C-means clustering algorithms and behave well in segmenting images of low signal-to-noise ratio.