A new method of SAR image segmentation based on fuzzy ART

For speckle consisted in SAR image, good segmentation result can't be obtained with traditional methods. In the paper, according to the features of SAR image and the fuzzy ART (adaptive resonance theory) neural network, a new gray level threshold method of SAR image segmentation is proposed, and the method can determine automatically the number of classes and suppress effectively the noise in the given SAR image. With experiments, it is proved that a good segmentation result can be obtained with this new method in the paper, and at the same time, the speckle in the SAR image is also reduced greatly.