Quantum Kernel Methods for InSAR Phase Unwrapping

Abstract

Phase unwrapping is the reconstruction of an absolute phase given its values modulo $2\pi$. It is a commonly applied post-processing technique in synthetic aperture radar interferometry applications, such as topographic height and deformation estimation. Established InSAR phase unwrapping approaches focus on large scene phase retrieval, aiming for good overall performance and relying on auxiliary data, such as additional acquisitions, to correct the resulting large local deviations. Given recent progress in quantum algorithm development for optimization problems and phase unwrapping, we're further exploring the feasibility of leveraging gate-based quantum computers and hybrid quantum algorithms to improve small-scale high-accuracy phase unwrapping. We consider a new quantum approach to $L^1$-norm phase unwrapping and a multi-stage combined approach leveraging the potential advantages of $L^0$ quantum and $L^1$ phase unwrapping.