Novel Fuzzy C-Means Segmentation Algorithm for Image with the Spatial Neighborhoods

The fuzzy c-means (FCM) algorithm is one of the most widely used method for data clustering, the standard FCM is not effective by itself to segment the image, as it fails to deal with the significant property of images, such as noise and intensity inhomogeneity. In this paper, we propose a novel fuzzy c-means image segmentation algorithm. Its effectiveness is due to two mechanisms. The first mechanism is the replacement of the Euclidean distance traditionally used to measure similarity of the image pixels by a novel similarity measure which is considered spatial neighborhoods using Gaussian kernel, and thus our method becomes less sensitive to the noise of the image. The second mechanism is not requirement of any similarity penalty term in FCM´s objective function as some FCM´s variants to reduce the influence of noise on the result of image segmentation, in addition, our method needs no requirement of setting parameter according to the image, and thus our method is more general and robust for image segmentation. We present the experimental results not only in the application of synthetic image segmentation, but also in the difficult application of synthetic aperture radar (SAR) image segmentation; it is shown that we propose a novel method to obtain the cluster for image segmentation.