Satellite image segmentation based on differential evolution

In this paper, an approach for segmentation is proposed for segmentation of satellite images. The proposed algorithm performs efficient segmentation by considering various aspects of satellite images like Hughes phenomenon, high correlation between spectral bands, etc. The proposed approach is a clustering based segmentation method that partitions image into different regions which represents land cover map. It uses type 2 fuzzy systems and differential evolution for producing accurate and precise segments. The proposed approach algorithm requires number of clusters to be specified beforehand. The proposed algorithm is validated by comparing the value of cluster validity indexes, silhouette index with well-known algorithms.