A Regularized Coherence Matrix Estimation Method for Phase Linking in Distributed Scatterer Interferometry

Changjun Zhao,Hanwen Yu,Yong Wang

DOI: https://doi.org/10.1109/igarss53475.2024.10641808

2024-01-01

Abstract:Phase linking is a key step in distributed scatterer interferometry (DSI), which can significantly reduce decorrelation by retrieving a consistent phase series. The performance of phase linking can be severely degraded by the inaccurate coherence magnitude matrix. Recently, some studies proposed to mitigate the problem by employing the regularization methods, e.g., adding a quantity to the diagonal or shrinking to the identity matrix. However, the correction is insufficient due to the simple structure assumption. In this study, we propose a new phase linking approach based on a powerful regularization method. Specifically, it achieves the maximum likelihood estimation of the coherence matrix under the structural constraint of total positivity. A simulated stack is exploited to test the performance of the proposed approach. The qualitative and quantitative evaluations demonstrate its superiority over the existing regularization methods.