A generic fundus image enhancement network boosted by frequency self-supervised representation learning
Abstract: Highlights•A generic network (GFE-Net) is proposed for fundus image enhancement.•Robust features are captured by self-supervised learning using frequency information.•Representation learning and image enhancement are coupled to boost the deployments.•Experiments validate the generalizability to unknown degradations in various data.•GFE-Net excels in comparison of image enhancement, efficiency, and generalizability.
Loading