Abstract:
Coherence estimation is crucial in interferometric synthetic aperture radar (InSAR) for various applications, including land cover classification, change detection, and m...Show MoreMetadata
Abstract:
Coherence estimation is crucial in interferometric synthetic aperture radar (InSAR) for various applications, including land cover classification, change detection, and multitemporal InSAR techniques. However, estimation challenges often arise due to systematic phases caused by deformation and topography, leading to an underestimation. The existing fast Fourier transform (FFT)-based stripe correction methods have limitations in handling complex patterns and are noise-sensitive. We propose here a novel approach using local phase surface fitting to remove the trend component and thereby improve the accuracy of the coherence estimation. The algorithm involves adaptive window size selection based on varying phase patterns and joint estimation of surface model parameters using regional network adjustment. Comparative experiments with simulated and Sentinel-1A data demonstrate the superiority of the method in effectively removing trend components, especially from nonlinear stripe regions. This improvement is evident in the significantly enhanced average coherence, with increases of 37.4% and 60.1% observed in the two experimental areas, resulting in enhanced quality of interferogram coherence maps.
Published in: IEEE Geoscience and Remote Sensing Letters ( Volume: 20)