Fringe enhanced phase unwrapping method based on unscented Kalman filter

Phase unwrapping is a crucial step in almost all Interferometric Synthetic Aperture Radar (InSAR) applications. However, the presence of noise often introduces phase inconsistencies, known as residues, which lead to ambiguous solutions. This paper builds upon the unscented Kalman filter phase unwrapping method and combines a modified phase gradient estimation algorithm, a pre-filtering step based on a neural network, and previous work to achieve higher precision in unwrapping the phase. The pre-filtering step aims to obtain a fringe-enhanced interferogram. The proposed method estimates the phase gradient of the pre-filtered interference phase and integrates it with the unscented Kalman filter phase unwrapping method. Additionally, it incorporates a strategy based on pixel reliability grades and introduces a new path-following strategy to obtain more accurate unwrapped phases. Unlike the traditional unscented Kalman filter phase unwrapping method, the proposed method can simultaneously unwrap multiple pixels, thereby improving the efficiency of the phase unwrapping process. Experimental results of simulated and real data demonstrate that the method proposed in this paper achieves better unwrapped results, especially when dealing with interferograms containing unclear fringes.