Transition Graphs for SAR Image Texture Characterization: An Exploratory Study

Abstract

We investigate the use of ordinal pattern transition graphs in the characterization of PolSAR image textures. Chagas et al. [4] proposed WATG, a transition graph analysis which includes the amplitude information of texture samples in the graph edges. This approach led to good results in the characterization and classification of data from urban, oceanic, and vegetation regions. In this work, we study how this approach behaves when considering the three intensity channels of PolSAR data, and its behavior when the samples are not homogeneous.