Robust Image Denoising of No-Flash Images Guided by Consistent Flash Images

Images taken in low light conditions typically contain distracting noise, and eliminating such noise is a crucial computer vision problem. Additional photos captured with camera flash can guide an image denoiser to preserve edges since the images often fine details reduced noise. Nonetheless, be misled by inconsistent images, which have structures (e.g., edges) that do not exist no-flash images. Unfortunately, this disparity frequently occurs as flash/no-flash pairs are different conditions. We propose learning-based technique robustly fuses while considering their inconsistency. Our framework infers consistent patches locally, similar ground truth, denoises using inferred ones via combination model. demonstrate our produce more robust results than state-of-the-art methods, given various structures. The source code available at https://github.com/CGLab-GIST/RIDFnF.