Unsupervised segmentation of HR SAR time series amplitude imagery aiming on context based change categorization

The analysis of recent spaceborne remote sensing images mainly implies dealing with high resolution (HR) imagery. Object-based analysis methods using segmentation results are well-suited for being applied on these images. Furthermore, neighborhood relations between the segments and shape-based features can be used to model the image content. In this paper, an unsupervised method for the segmentation of HR SAR time series amplitude images is proposed. This method represents the pre-processing of following investigations aiming on the context-based categorization of detected changes in SAR time series data.