OSMamba: Omnidirectional Spectral Mamba with Dual-Domain Prior Generator for Exposure Correction

TL;DR

OSMamba is a novel exposure correction network that leverages omnidirectional spectral scanning and a dual-domain prior generator, integrating frequency domain modeling and generative diffusion to improve detail restoration in challenging real-world scenarios.

Contribution

The paper introduces OSMamba, combining frequency domain long-range dependency modeling with a diffusion-based prior generator for enhanced exposure correction.

Findings

Achieves state-of-the-art results on multiple-exposure datasets.

Effectively restores details in severely under- and over-exposed regions.

Outperforms existing methods both quantitatively and qualitatively.

Abstract

Exposure correction is a fundamental problem in computer vision and image processing. Recently, frequency domain-based methods have achieved impressive improvement, yet they still struggle with complex real-world scenarios under extreme exposure conditions. This is due to the local convolutional receptive fields failing to model long-range dependencies in the spectrum, and the non-generative learning paradigm being inadequate for retrieving lost details from severely degraded regions. In this paper, we propose Omnidirectional Spectral Mamba (OSMamba), a novel exposure correction network that incorporates the advantages of state space models and generative diffusion models to address these limitations. Specifically, OSMamba introduces an omnidirectional spectral scanning mechanism that adapts Mamba to the frequency domain to capture comprehensive long-range dependencies in both the amplitude and phase spectra of deep image features, hence enhancing illumination correction and structure recovery. Furthermore, we develop a dual-domain prior generator that learns from well-exposed images to generate a degradation-free diffusion prior containing correct information about severely under- and over-exposed regions for better detail restoration. Extensive experiments on multiple-exposure and mixed-exposure datasets demonstrate that the proposed OSMamba achieves state-of-the-art performance both quantitatively and qualitatively.