Urban SAR Tomography With <italic>Z</italic>-Structure Constraint of Buildings

Abstract

Synthetic aperture radar tomography (TomoSAR) technology effectively mitigates issues such as severe layover in high-resolution synthetic aperture radar (SAR) urban imagery, presenting numerous applications in reconstructing complex 3-D urban scenes. However, prevailing TomoSAR methodologies usually overlook the geometrical feature of the targets, which are particularly pronounced in buildings in urban scenes. To improve the quality of 3-D reconstruction of building targets for large-scale urban scenes, we propose geometric-guided TomoSAR processing in this article. First, the geometrical feature referred to a Z-structure of buildings in TomoSAR point cloud is studied. Then, a Z-structure information extraction approach is proposed to provide prior information for subsequent geometric constraints. Finally, the Z-structure-constraint-based tomographic algorithm is proposed, optimizing the solution space of the original algorithms. Experiments are conducted on real SAR data, and 3-D point clouds of the entire urban scenes are obtained. The proposed algorithm demonstrates commendable performance in point cloud entropy and concentration, especially in the inversion of different floor structures.

Keywords

• Compressed sensing (CS)
• geometrical feature
• synthetic aperture radar tomography (TomoSAR)
• 3-D reconstruction of buildings