Summary
Improving low-light photos is more complicated than just making them brighter because shadows, noise, and color issues vary across the image. The authors propose a new way to add illumination information safely into a special type of image model called an SSM via a small add-on called a bottleneck-side PromptAdapter. This approach predicts a map of lighting and uses it to guide the model without changing its core structure. Tests on standard low-light image datasets showed consistent improvements in image quality metrics when using this method. The authors caution that how the conditioning is designed is as important as the lighting information itself.
What this means in practice
- •For mobile app developers: Enhance photo apps with improved low-light enhancement that more reliably handles noise and color issues using safe illumination guidance in SSM-based models.
- •For camera firmware engineers: Incorporate stable illumination conditioning techniques into embedded low-light image processing pipelines to improve output quality without redesigning core algorithms.
Authors
Wei Chen Ng, Qiyi Xie, Jian Ding Tan, Yan Kang Lee, Abdulrab Hakim Qaid Abdullah, Wei Hown Tee
Abstract
Low-light image enhancement (LLIE) requires more than brightness amplification because real low-light images often exhibit spatially varying illumination, amplified noise, color distortion, and local detail loss. Recent SSM- and Mamba-based restorers provide efficient long-range modeling, yet the route for introducing illumination priors into such backbones remains insufficiently studied. This study addresses safe illumination conditioning for SSM-based LLIE through a bottleneck-side PromptAdapter. A lightweight illumination estimator first predicts a single-channel illumination map, which is then converted into prompt tokens and injected through a residual sidecar adapter at the bottleneck, thereby avoiding direct modification of the state-space core. Under a locked three-seed evaluation protocol on LOL-v1, LOLv2-Real, and LOLv2-Synthetic, the PromptAdapter improves over the Base model by 0.94 dB, 1.24 dB, and 0.93 dB in PSNR, respectively, while also improving SSIM, LPIPS, and masked dark-region metrics. The clearest real-data evidence appears on LOLv2-Real. A Learned-Fixed control on LOL-v1 indicates that the present results support the utility of the sidecar conditioning pathway more strongly than a strict claim of illumination-conditioned superiority, and the LOLv2-Synthetic result remains stability-conditioned. These findings suggest that, in SSM-based LLIE, conditioning design can be as important as the conditioning signal itself.