Optimization of PSInSAR networks with application to TomoSAR for full detection of single and double persistent scatterers

In this paper, we develop an optimized network to achieve the full detection of single and double persistent scatterers (PSs) in SAR tomography (TomoSAR). The two-tier network used in persistent scatterer interferometry (PSInSAR) is applied to TomoSAR. Adaptive network densification is constructed to improve global connectivity in the first-tier network. This optimization contributes to improving the connectivity of different clusters by sacrificing computational efficiency slightly. In the second-tier network, we develop an omnidirectional point extension method by progressively constructing local networks and multi-directional extension. This contributes to identifying scatterers that are far away from the reference points in the first-tier network. This improved point extension method can achieve the full extraction of single and double PSs, which is especially applicable in areas with unevenly and sparsely distributed PSs. To validate these improvements, we applied the method to the western reclaimed area of Shenzhen, China with offshore islands and facilities using 30 COSMO-SkyMed images.