Two-stage AI trick can finally fix the low-light struggle of smartphone cameras
Researchers have developed a two-stage AI system that enhances ultra-high-definition low-light images while preserving fine textures, edges and scene details.
Low-light photography on smartphones has always been a challenge, with grainy images, muddy shadows, and lost fine details plaguing even expensive devices. A new AI-based image enhancement technique called LL-Refiner from researchers at Wuhan University in China may offer a solution. This two-stage machine-learning system aims to improve ultra-high-definition low-light images without sacrificing the natural look.
The first stage uses a Transformer-based neural network on a lower-resolution version of the photo, focusing on global characteristics like illumination and color distribution. The second stage, an adaptive refinement network, uses cross-attention modules to progressively restore sharper edges, textures, and fine details, gradually increasing the image's resolution.
Tested against other enhancement methods, LL-Refiner consistently produced better results, particularly in maintaining textured areas and fine structures. Importantly, it also improved depth estimation, a crucial factor for cameras feeding visual information into technologies like robots and autonomous vehicles. However, this research is still a demonstration rather than a commercial feature.
As smartphone cameras continue to push toward higher resolutions, AI techniques like LL-Refiner could be the key to intelligently reconstructing difficult low-light images without compromising their authenticity.
Written by urgent.news from Digital Trends's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
