Provably Convergent Plug-and-Play Quasi-Newton Methods

TL;DR

This paper introduces a provably convergent plug-and-play quasi-Newton method that accelerates convergence in inverse imaging problems while maintaining minimal assumptions on the denoiser.

Contribution

It proposes a novel quasi-Newton approach within the provable PnP framework, enabling faster convergence with weaker denoiser assumptions.

Findings

Achieves 2-8x faster convergence in image deblurring and super-resolution.

Retains convergence guarantees under weak assumptions on the denoiser.

Demonstrates effectiveness on inverse imaging tasks.

Abstract

Plug-and-Play (PnP) methods are a class of efficient iterative methods that aim to combine data fidelity terms and deep denoisers using classical optimization algorithms, such as ISTA or ADMM, with applications in inverse problems and imaging. Provable PnP methods are a subclass of PnP methods with convergence guarantees, such as fixed point convergence or convergence to critical points of some energy function. Many existing provable PnP methods impose heavy restrictions on the denoiser or fidelity function, such as non-expansiveness or strict convexity, respectively. In this work, we propose a novel algorithmic approach incorporating quasi-Newton steps into a provable PnP framework based on proximal denoisers, resulting in greatly accelerated convergence while retaining light assumptions on the denoiser. By characterizing the denoiser as the proximal operator of a weakly convex function, we show that the fixed points of the proposed quasi-Newton PnP algorithm are critical points of a weakly convex function. Numerical experiments on image deblurring and super-resolution demonstrate 2--8x faster convergence as compared to other provable PnP methods with similar reconstruction quality.