Equivariant Learning for Unsupervised Image Dehazing

TL;DR

This paper introduces an unsupervised equivariant learning framework for image dehazing that leverages symmetry and haze physics, outperforming existing methods especially in scientific imaging contexts.

Contribution

The paper presents a novel unsupervised equivariant learning approach for image dehazing that models haze physics and enforces symmetry, improving performance without requiring paired data.

Findings

Outperforms state-of-the-art dehazing methods on scientific and natural images

Effectively models haze physics through adversarial learning

Enables haze removal without ground truth images

Abstract

Image Dehazing (ID) aims to produce a clear image from an observation contaminated by haze. Current ID methods typically rely on carefully crafted priors or extensive haze-free ground truth, both of which are expensive or impractical to acquire, particularly in the context of scientific imaging. We propose a new unsupervised learning framework called Equivariant Image Dehazing (EID) that exploits the symmetry of image signals to restore clarity to hazy observations. By enforcing haze consistency and systematic equivariance, EID can recover clear patterns directly from raw, hazy images. Additionally, we propose an adversarial learning strategy to model unknown haze physics and facilitate EID learning. Experiments on two scientific image dehazing benchmarks (including cell microscopy and medical endoscopy) and on natural image dehazing have demonstrated that EID significantly outperforms state-of-the-art approaches. By unifying equivariant learning with modelling haze physics, we hope that EID will enable more versatile and effective haze removal in scientific imaging. Code and datasets will be published.