A novel Bidirectional Cross-frequency Domain Interaction Network (BiFreqNet) for UHD LLIE is proposed, which incorporates a Dual-domain Feature Interaction module into the multi-scale representations of a pretrained variational autoencoder to perform reciprocal enhancement between low- and high-frequency features.
Abstract
Existing Ultra-High-Definition (UHD) low-light image enhancement (LLIE) methods typically rely on either spatial- or frequency-domain modeling, or independently process different frequency components for efficient restoration. However, they often fail to jointly exploit complementary spatial--frequency information and the inherent relationship between illumination and structural information across frequencies, leading to inconsistent illumination, residual noise, and detail artifacts. In this paper, we propose a novel Bidirectional Cross-frequency Domain Interaction Network (BiFreqNet) for UHD LLIE. Specifically, BiFreqNet incorporates a Dual-domain Feature Interaction (DFI) module into the multi-scale representations of a pretrained variational autoencoder (VAE) to perform reciprocal enhancement between low- and high-frequency features. DFI consists of a Spatial and Spectral-Aware Low-Frequency Enhancement (S$^2$LFE) block and a Cross-Frequency Enhancement (CFE) block. S$^2$LFE exploits high-frequency structural cues to guide low-frequency restoration and jointly models spatial and spectral information to recover global exposure and spatially varying illumination while suppressing sensor noise. CFE further leverages the enhanced low-frequency contextual information to guide adaptive high-frequency refinement, enabling faithful reconstruction of structures and fine details. Extensive experiments on four general and UHD LLIE benchmarks demonstrate that BiFreqNet achieves competitive or superior performance compared with state-of-the-art methods.
The Spatial-Frequency Fusion Network (SFF-Net), which decouples intensity and chroma in HVI space and then learns frequency-aware enhancement with cross-branch compensation, is proposed, which improves SICE, Sony-Total-Dark, and unpaired real image results with only 1.931 M parameters.
Ying-Bo Jia· Conference on Advanced Algor...· 0 citations
Low-light image enhancement remains challenging due to the spatially heterogeneous nature of degradation, as different regions within the same image often require substantially different degrees of illumination correction and detail restoration. However, existing methods typically adopt spatially uniform processing, wh...
Hao-Dian Wang· Proceedings of the 34th ACM...· 0 citations
Although the multimode fiber (MMF) imaging enables ultra-thin minimally invasive probes, strong modal coupling and fiber perturbations introduce severe distortions that hinder accurate image reconstruction. This paper proposes a frequency-aware dual-domain reconstruction framework, termed as the spatial–frequency domai...
Ji Liang, Zheyi Yao, Chun-Cheng Zhang et al.· Journal of Lightwave Technol...· 0 citations
Frequency-domain methods have shown strong potential for low-light image enhancement (LLIE) because they can effectively regulate global illumination. However, most existing methods assume that amplitude and phase can be restored independently, while overlooking their inherent spectral coupling. This artificial decoupl...
Tong-Shun Zhang, Zi-Xuan Zhong, Ping-Ping Liu et al.· 0 citations
Underwater image enhancement (UIE) is still difficult. This is mainly due to wavelength-dependent light absorption, scattering, and environmental noise, which lead to low visibility, color distortion, and loss of fine details. Most existing deep learning (DL) based models primarily operate on spatial-domain convolu...
Najmul Hassan, Munsif Ali, Abu Saleh Musa Miah et al.· Scientific Reports· 0 citations
Low-light image enhancement is challenging due to noise and poor visibility. While diffusion models have shown remarkable generative capabilities, applying them directly to this task in the pixel domain remains problematic. They tend to struggle with the trade-off between global illumination correction and local detail...
Ping-Ping Liu, Jia-Hao Liu, S. Pei et al.· Neural Networks· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.