Local and spatial information based fuzzy C-Means clustering for color image segmentation

Fuzzy C-Means (FCM) algorithm is the most popular method used in image segmentation for clustering because it has robust characteristics for ambiguity and can retain much more information than hard segmentation methods. Although the conventional FCM algorithm works well on most noise-free images, it is very sensitive to noise and other imaging artifacts, since it does not consider any information about spatial context. We propose a new method for image clustering by variation of traditional fuzzy c-means algorithm to provide noise free extracted image. In the new method local spatial information and gray level information is incorporated into it. The new method is called FLICM and it is fully free of empirically adjusted parameters and provides robustness to noisy images. RGB image is converted to gray intensity image by HSI model. The gray image is clustered using FLICM algorithm and it is converted again to segmented color image. Segmentation in intensity band considerably reduces the time of processing instead of processing the three bands of the RGB model and it results in high degree of accuracy.