Unsupervised PolSAR Change Detection Based on Polarimetric Distance Measurements and ConvLSTM Network

Abstract

Time-series PolSAR are capable for continuous change monitoring of natural resources and urban land-covers regardless of weather and lighting conditions. However, in the big SAR data era, the scarcity of labeled PolSAR samples poses new challenge to the traditional change detection methods. To reduce the dependence on labeled samples and ensure the efficiency of long time-series PolSAR interpretation, an unsupervised and pseudolabel-based change detection method is proposed. First, the similarity maps of time-series PolSAR are gauged by three selected polarimetric distance measurements (PDMs), which are suitable for PolSAR distribution characteristics and have the potential to reflect PolSAR changes. Second, the high-confidence changed pseudosamples are selected based on the similarity maps, and the unchanged pseudosamples are selected based on the nonsimilarity maps. Third, the limited selected pseudosamples (changed and unchanged) and multidimensional features are used to train the ConvLSTM network for change detection, and the input features include the T3 coherence matrix elements of time-series PolSAR and the aforementioned PDMs. Finally, the change detection results based on pseudosamples and the ConvLSTM network can be obtained, without additional manual labels. Adequate experiments are conducted on Radarsat-2, UAVSAR full-polarized, and Sentinel-1 dual-polarized datasets, achieving improved unsupervised change detection accuracy at 89.59–93.24%.

Keywords

• Long short-term memory (LSTM) network
• polarimetric distance measurements (PDMs)
• synthetic aperture radar (SAR)
• time series images
• unsupervised change detection