Three-Dimensional Reconstruction of Buildings IN Multi-Look Tomosar Based on Homogeneous Pixels and GLRT

Due to the distinctive side-looking imaging characteristics of Synthetic Aperture Radar (SAR), numerous geometric distortions exist, such as layover, in complex urban scenes, making it challenging to meet interpretation requirements. Synthetic Aperture Radar Tomography (TomoSAR), as a genuine three-dimensional imaging technique, can distinguish discrete targets within the same pixel, and demonstrate significant potential in urban building monitoring. This paper proposes a multi-look tomography method based on homogeneous pixels and a Generalized Likelihood Ratio Test (GLRT), to separate and calculate the height information of scatterers compressed within the same pixel. Experiments were conducted using a dataset from TerraSAR-X in the Beijing region. The results indicate that the proposed method can effectively separate single and multiple scatterers, with a computed building height error not exceeding 1 meter. Additionally, the method is characterized by its simplicity of operation and high processing speed.