Enhanced Phase Optimization Using Spectral Radius Constraints and Weighted Eigenvalue Decomposition for Distributed Scatterer InSAR

Eigenvalue decomposition (EVD) of covariance matrices or coherence matrices has been employed to suppress noise in phase information, and this approach has shown some effectiveness in data processing. However, while this method helps attenuate noisy phase components, it also tends to significantly degrade the true deformation phase information, which can be detrimental in certain applications. To address this issue, this paper proposes an optimal eigenvalue decomposition phase optimization method, incorporating a spectral radius-constrained covariance matrix construction, named SREVD. This method constructs a covariance matrix using spectral radius constraints and then selects optimal eigenvectors from the covariance matrix for weighted combination, yielding the final optimized phase. The advantages of this approach (1) include the use of spectral radius constraints to obtain a stable covariance matrix, and (2) rather than using the eigenvector associated with the maximum eigenvalue for phase optimization, the interferometric phase is reconstructed by a weighted combination of eigenvectors selected through eigenvalue-based optimization. Experimental analysis conducted in a mining area in Datong, Shanxi Province, China, yields the following conclusions: compared to the original interferogram and the traditional EVD-optimized interferogram, the proposed SREVD method demonstrates superior noise suppression. After optimization with SREVD, the density of monitoring points has been significantly improved. The final number of selected points is 9.06 times that of StaMPS and 1.3 times that of EVD optimization, which can better reflect the topographic changes in the study area.