Skip to Main Content
We propose an adaptive complex-amplitude texture classifier using local phase unwrapping and complex-valued feature extraction for interferometric SAR images. We choose complex-valued mean and second-order covariances with slope-direction insensitivity as the feature vector that represent the statistical characteristics. We also introduce local phase unwrapping method to handle the widely distributed singular points. We found that interferometric SAR images can be classified successfully in clusters corresponding to a lake, a mountain, and so on.