Frequent human activity associated with rapid urbanization may lead to land subsidence and instability of buildings. High-resolution synthetic aperture radar (SAR) interferometry plays a significant role in urban monitoring. However, single-track interferometric synthetic aperture radar (InSAR) technology is limited to capturing only one side of a building, necessitating the fusion of multi-track InSAR observations. Existing methods for fusing multi-track InSAR point clouds struggle to extract sufficient identical features and rely on external data. This paper proposes an InSAR point cloud registration strategy to efficiently fuse multi-source data of buildings using the quadratic density-based spatial clustering of applications with noise (DBSCAN) clustering and multi-source feature point matching. The ascending and descending TerraSAR-X images are used to retrieve the surface deformation and building height of Fuzhou City. The point cloud registration experiments demonstrate that the height estimation accuracy of the InSAR point clouds approaches 2 m, and the insufficient and uneven feature matching tends to registration error or local optimum. The evenly distributed point pairs with numbers greater than 20 are better for accurate alignment of point clouds. Moreover, the multi-dimensional deformations of the buildings are retrieved based on the fused point clouds, with emphasis on assessing the health condition of two buildings. The deformation results reveal that settlements occur in low architectural complex with the maximum deformation velocity of 40 mm/y since the existence of soft soil, high-density population, and unstable foundation, while high-rise buildings behave relatively stable.
Wu et al. (Fri,) studied this question.