A new automatic segmentation for synthetic aperture radar images

The multiplicative nature of the speckle noise in SAR images has been a big problem in SAR image segmentation. A novel method for automatic segmentation of SAR images is proposed. Firstly, we use wavelet energy to extract texture features, use regional statistics to extract gray-level features and use edge preserving mean of gray-level features to ensure the accuracy of classification of pixels near to the edge. Three representative kinds of features of SAR image are extracted, so the segmentation ability is enhanced. Then an improved unsupervised clustering algorithm is proposed for image segmentation, which can determine the number of classes automatically. Segmentation results on a real SAR image demonstrate the effectiveness of the proposed method.