Abstract: Highlights•A novel Laplacian pyramid-based network (LPUDC) is proposed for UDC image restoration.•LPUDC decomposes images into high- and low-frequency branches for targeted degradation handling.•For the first time in UDC image restoration, we introduce a Deep Guided Filtering (DGF) module to fuse multi-scale features with low computational cost.•The proposed method achieves superior results on two widely used UDC benchmark datasets.
External IDs:dblp:journals/ijon/DingFWDZLYML25
Loading