In this paper we proposed an improved color based K-mean algorithm for clustering of satellite Image (SAR). Image clustering is the versatile method that can help in the provocative task of efficient search in very fast-growing image data bases. It plays an important role in image analysis, pattern recognition, image segmentation etc. Our method comprises of two stages. The first step is the calculation of initial cluster centres using interactive selection process. In the second step these points are given as input to K-mean clustering algorithm that clusters the image based on color and Minimum Square Euclidean distance. We have evaluated results using Dunns index and Davis bound index.