Distributed Scatterer Interferometry From Traditional Methods to Deep Learning: A comprehensive review

Interferometric synthetic aperture radar (InSAR) enables high-precision measurement of Earth surface deformation, providing essential support for geohazard monitoring and geophysical parameter retrieval. However, long-term monitoring is challenged by temporal decorrelation, which can significantly degrade the signal-to-noise ratio (SNR) of the interferometric phase. To overcome this challenge, distributed scatterer (DS) interferometry (DSI) reconstructs a temporally consistent phase series by leveraging redundant interferograms. Over the past 15 years, traditional model-based methods have undergone steady refinement, while the recent introduction of deep learning techniques into this field has injected new vitality. This article provides a comprehensive review of both traditional model-based and emerging deep learning-driven approaches. Specifically, we review the complete DSI processing workflow, including homogeneous pixel selection, complex coherence estimation, interferometric phase series estimation, and DS identification. For each stage, representative methods from both traditional and learning-based approaches are analyzed, supported by simulated and real-data experiments. Finally, we summarize the strengths and limitations of existing techniques, identify current challenges, and outline promising directions for future research. This article aims to provide a comprehensive overview of the current state of DSI and to inspire further research in this vibrant and rapidly evolving field.