Spatio-Temporal Adaptive Fusion Network Based on Spatial-Channel-Axis Attention Mechanism

Spatio-temporal fusion of remote sensing imagery aims to generate images with both high spatial and temporal resolution by integrating the advantages of low-spatial high-temporal and high-spatial low-temporal resolution imagery. However, existing methods face challenges in handling complex surface changes and multi-scale feature extraction. This paper proposes a novel Spatio-Temporal Adaptive Fusion Network based on Spatial-Channel-Axis Attention Mechanism (SCAP-STFNet), which enhances fusion accuracy via a newly designed SCAP module and gating mechanism. The network adopts an encoder-decoder structure, where the encoder leverages the SCAP module for multi-dimensional attention feature extraction, and the decoder performs adaptive feature fusion through gating. Experiments on CIA and LGC datasets demonstrate that the proposed method outperforms mainstream approaches in terms of RMSE, SSIM, and UIQI, verifying its effectiveness and superiority.