Synthetic aperture radar tomography (TomoSAR) emerges as an advanced interferometric SAR (InSAR) technique for 3D imaging as well as deformation monitoring. The state-of-the-art TomoSAR algorithms harness the capabilities of compressive sensing (CS)-based sparse reconstruction, showing unprecedented super-resolution power and location accuracy. However, the computational demands of CS-based TomoSAR algorithms render them impractical for large-scale processing. Addressing this challenge, this dissertation develops novel deep learning-based algorithms tailored for efficient and accurate super-resolution TomoSAR inversion in large-scale processing.
«
Synthetic aperture radar tomography (TomoSAR) emerges as an advanced interferometric SAR (InSAR) technique for 3D imaging as well as deformation monitoring. The state-of-the-art TomoSAR algorithms harness the capabilities of compressive sensing (CS)-based sparse reconstruction, showing unprecedented super-resolution power and location accuracy. However, the computational demands of CS-based TomoSAR algorithms render them impractical for large-scale processing. Addressing this challenge, this dis...
»