Holistic Dynamic Frequency Transformer for Image Fusion and Exposure Correction

TL;DR

This work tackles exposure-related image restoration by unifying low-light enhancement, exposure correction, and multi-exposure fusion within a frequency-domain framework. It introduces Holistic Frequency Attention and Dynamic Frequency Feed-Forward Network to form a Holistic Dynamic Frequency Transformer, paired with Laplacian-pyramid decomposition and pyramid fusion to enable frequency-specific restoration across multiple scales. Key contributions include a unified architecture for exposure tasks, a frequency-domain attention mechanism with $O(N \log N)$ complexity, and extensive ablations and benchmarks showing state-of-the-art performance on exposure correction, low-light enhancement, MEF, and improvements on downstream high-level tasks when used as a pre-enhancer. The approach demonstrates robust, high-quality restoration with reduced computational cost, offering a practical pathway to more versatile and unified exposure-correction solutions in real-world vision systems.

Abstract

The correction of exposure-related issues is a pivotal component in enhancing the quality of images, offering substantial implications for various computer vision tasks. Historically, most methodologies have predominantly utilized spatial domain recovery, offering limited consideration to the potentialities of the frequency domain. Additionally, there has been a lack of a unified perspective towards low-light enhancement, exposure correction, and multi-exposure fusion, complicating and impeding the optimization of image processing. In response to these challenges, this paper proposes a novel methodology that leverages the frequency domain to improve and unify the handling of exposure correction tasks. Our method introduces Holistic Frequency Attention and Dynamic Frequency Feed-Forward Network, which replace conventional correlation computation in the spatial-domain. They form a foundational building block that facilitates a U-shaped Holistic Dynamic Frequency Transformer as a filter to extract global information and dynamically select important frequency bands for image restoration. Complementing this, we employ a Laplacian pyramid to decompose images into distinct frequency bands, followed by multiple restorers, each tuned to recover specific frequency-band information. The pyramid fusion allows a more detailed and nuanced image restoration process. Ultimately, our structure unifies the three tasks of low-light enhancement, exposure correction, and multi-exposure fusion, enabling comprehensive treatment of all classical exposure errors. Benchmarking on mainstream datasets for these tasks, our proposed method achieves state-of-the-art results, paving the way for more sophisticated and unified solutions in exposure correction.

Figures (11)

• Figure 1: The proposed method produce a single-exposure-corrected(SEC)/multiple-exposure-fused(MEF) image with clear details and visually pleasing colors.
• Figure 2: The workflow of the proposed Restorer.
• Figure 3: The workflow of the proposed Holistic Dynamic Frequency Transformer.
• Figure 4: Qualitative Comparison of exposure correction performance on MSEC dataset.
• Figure 5: Qualitative Comparison of exposure correction performance on LCDP dataset.