Flat Coherence Metric Based InSAR Phase Estimation for Distributed Scatterers

State-of-the-art methods in phase linking (PL) for distributed scatterer interferometry (DSI) extract consistent phase histories from coherence matrices, either based on sample matrices or those with calibrated magnitudes. However, the reliability of sample coherence as an estimator diminishes in scenarios involving large-dimensional data. To address this limitation, we introduce a novel PL approach, termed LaMIE, specifically designed for precise phase history retrieval from large-dimensional coherence matrices in DSI. In this context, "largedimensional" denotes instances where the temporal dimension N of coherence matrices aligns with the number P of statistically homogeneous pixels (SHP). LaMIE comprises two key steps: 1) shrinking the sample coherence matrix, and 2) executing phase history retrieval through the flat coherence metric. Our proposed method's efficacy is rigorously validated through experiments using both simulated and real data, demonstrating superior performance when compared to existing PL techniques. Through the implementation of LaMIE, the densities of selected points with stable phases witness significant improvement, allowing for the retrieval of displacement velocities in more regions than achievable with current SOTA methods.