Cell Image Segmentation Based on Color Mutual Information

The selection of the clustering parameter based on k-means plays an important part in the cell image segmentation. By combination different clustered image´s color information entropy calculation with original image, it can gain the optimal clustering number for color cell image segmentation. It also introduces a clustering number k-related mutual information and mutual information error calculation to increase algorithm´s robustness and can get a better sole clustering result automatically. Experiments show that this method has better segmentation result and calculation efficiency.