Images shared on the Internet are often compressed into a small size, and thus have the JPEG artifact. This issue becomes challenging for task of low-light image enhancement, as the artifacts hidden in dark image regions can be further boosted by traditional enhancement models. We use a divide-and-conquer strategy to tackle this problem. Specifically, we decompose an input image into an illumination layer and a reflectance layer, which decouple the issues of low lightness and JPEG artifacts. Therefore we can deal with them separately with the off-the-shelf enhancing and deblocking techniques. Qualitative and quantitative comparisons validate the effectiveness of our method.
展开▼
