Low-Light Aerial Imaging With Color and Monochrome Cameras

Aerial imaging aims to produce well-exposed images with rich details. However, aerial photography may encounter low-light conditions during dusk or dawn, as well as on cloudy or foggy days. In such low-light scenarios, aerial images often suffer from issues such as underexposure, noise, and color distortion. Most existing low-light imaging methods struggle with achieving realistic exposure and retaining rich details. To address these issues, we propose an Aerial Low-light Imaging with Color-monochrome Engagement (ALICE), which employs a coarse-to-fine strategy to correct low-light aerial degradation. First, we introduce wavelet transform to design a perturbation corrector for coarse exposure recovery while preserving details. Second, inspired by the binocular low-light imaging mechanism of the human visual system, we introduce uniformly well-exposed monochrome images to guide a refinement restorer, processing luminance and chrominance branches separately for further improved reconstruction. Within this framework, we design a Reference-based Illumination Fusion Module (RIFM) and an Illumination Detail Transformation Module (IDTM) for targeted exposure and detail restoration. Third, we develop a Dual-camera Low-light Aerial Imaging (DuLAI) dataset to evaluate our proposed ALICE. Extensive qualitative and quantitative experiments demonstrate the effectiveness of our ALICE, achieving a PSNR improvement of at least 19.52% over 12 state-of-the-art methods on the DuLAI Syn-R1440 dataset, while providing more balanced exposure and richer details. Our codes and datasets will be made publicly available after the peer review process.