Abstract:

Fuzzy clustering method can get higher classification accuracy than the hard clustering, but it still relies on the prior assumptions on the number of clusters. This paper proposes an automatic fuzzy clustering method based on self-adaptive differential evolution for remote sensing image (AFCDE). The proposed AFCDE algorithm can adaptively find the optimal number of clusters and obtain the satisfied classification result based on Xie-Beni index by utilizing the fast, robust and efficient global search algorithm, differential evolution (DE) algorithm. Three experimental results with one simulated image and two real remote sensing images show that the proposed algorithm not only finds the optimal number of clusters, but also outperforms the traditional clustering algorithms, such as K-means, ISODATA and fuzzy K-means.

Key words: remote sensing, differential evolution, fuzzy clustering, automatic clustering