Cross-Level Interaction and Intralevel Fusion Network for Remote Sensing Image Dehazing

Existing approaches have significantly advanced remote sensing image dehazing. However, they often rely on conventional encoder-decoder architectures, leading to prolonged inference times. To this end, we propose a novel cross-level interaction and intralevel fusion network (CINet) for remote sensing image dehazing, which shifts the focus from encoder-decoder dependencies to an innovative hierarchical architecture centered on skip connections, leading to competitive dehazing performance with decreased calculating complexity. Furthermore, we introduce a cross-level multiscale interaction module (CMIM) to facilitate effective interactions between features across hierarchical levels, mitigating the information loss commonly caused by repeated downsampling operations. Meanwhile, we develop an intralevel dual-dimension fusion module (IDFM), which leverages height-wise and width-wise self-attention to capture rich spatial-aware information, enabling robust and efficient intralevel feature fusion. Additionally, we propose a multiview progressive extraction block (MPEB), which decomposes features into four distinct components and applies convolutions with diverse kernel sizes, groups, and dilation factors. This design promotes progressive feature learning while significantly reducing computational overhead. Extensive experiments conducted on nine publicly available datasets validate the effectiveness and superiority of our proposed model.