Multi-Objective Fuzzy Clustering for Synthetic Aperture Radar Imagery

Synthetic aperture radar (SAR) images clustering has always been an important and challenging task. Due to the complexity of SAR images and the lack of relevant prior knowledge, the traditional clustering methods cannot work well in SAR images clustering. The appearance of multi-objective optimization clustering algorithms provides us a powerful tool for analyzing SAR images. The existing multi-objective clustering methods often take the energy function J_m of the fuzzy c-means (FCM) algorithms and the Xie-Beni (XB) index as two objective functions. However, these multi-objective clustering methods do not consider the spatial and contextual information, which can greatly improve the robustness to noise and outliers. Therefore, in this letter, we propose a multi-objective clustering algorithm which simultaneously optimizes both the energy function J_s of the fast generalized fuzzy c-means (FGFCM) and XB index. The proposed method enhances its robustness to noise and outliers by introducing local spatial and grey level information together through the energy function J_s. The experimental results on both the synthetic and real SAR images demonstrate the effectiveness and superiority of the proposed method.