Cloud Extraction of Satellite Image Using Fuzzy C-Means Clustering Approach

Abstract

We notice that each kind of image has its own feature and these features can recognize by human brain easily. The difference between person and computer is that person can notice the major interest object of a picture and ignore others. We propose a new segmentation method of satellite images based on fuzzy c-means clustering (FCM) to extract an object, cloud. Our method only promise extracts the cloud from a satellite image clearly just like the weather expert only notice the cloud. We extract R, G, and B components of a true color satellite image and convert a true color satellite image to a 256 gray image, from one image get four images. The ANOVA algorithm was used to analysis the cloud images and its results shown that only the discrepancy between the group of “R-components” and the group of “Gray images” are not significance. In other word, we can use the “R-components” to extract the cloud from a satellite image, instead of the converted gray image, and then the operation time could be shortened.