Learning Reference-Guided Exposure Correction with Hybrid Illumination Characteristics

2026-05-26

AI Summary

This work addresses the challenge of detail loss in image exposure correction under no-reference or weak-reference conditions by proposing HICNet, a novel framework that leverages a lightweight, content-agnostic encoder to extract mixed illumination features. The method drives a multi-scale modulation network through embedding discrepancies between source and reference images, integrating global FiLM modulation with photometric channel rebalancing for precise spectral control. Innovatively, it introduces illumination manifold contrastive learning, enabling end-to-end training without requiring ground-truth labels or intrinsic image decomposition. Experimental results demonstrate that HICNet achieves superior accuracy on public benchmarks and exhibits strong generalization capabilities on unseen scenes.

Abstract

We present HICNet, a reference-guided exposure correction framework. A lightweight, content-agnostic encoder distills each image into a compact illumination embedding capturing regional brightness, edge contrast, and higher-order luminance moments. The embedding difference between a source and its reference drives a multi-scale modulation network that combines FiLM-based global adjustment with Photometric Channel Rebalancing for fine-grained, illumination-aware spectral gating, producing exposure-matched outputs while faithfully preserving scene details. A cross-batch contrastive loss orders the illumination manifold, bolstering robustness to diverse lighting conditions. Trained without ground truth or intrinsic decomposition, HICNet attains better accuracy on public benchmarks and generalizes well to entirely unseen scenes.

Problem

• exposure correction
• reference-guided
• illumination characteristics
• image enhancement
• photometric consistency

Innovation

• reference-guided exposure correction
• illumination embedding
• multi-scale modulation
• photometric channel rebalancing
• contrastive illumination ordering