Overlapped Wavelet Diffusion for Low-Light Image Enhancement
Researchers have developed a new framework called Overlapped Wavelet Diffusion (OWDiff) to improve low-light image enhancement. This method addresses blocking artifacts common in previous diffusion-based techniques by using an Overlapped Wavelet Transform. Additionally, OWDiff incorporates a High-Frequency Enhance Block guided by low frequencies to recover finer details and textures, resulting in sharper images. Experiments show OWDiff outperforms existing methods on standard datasets, achieving better visual quality and quantitative metrics. AI
IMPACT Enhances image quality in low-light conditions, potentially improving AI vision systems' performance in such environments.