A Spatiotemporal Fusion Method Based on Multiscale Feature Extraction and Spatial Channel Attention Mechanism

Remote sensing satellite images with a high spatial and temporal resolution play crucial role in Earth science applications. However, due to technology cost constraints, it is difficult for single achieve both resolution. The spatiotemporal fusion method cost-effective solution generating dense data In recent years, image based on deep learning has received wide attention. this article, multiscale feature extraction channel attention mechanism proposed. Firstly, the uses fully utilize structural features images. Then novel used capture information; finally, rich information are fuse Experimental results obtained from two datasets show that proposed outperforms existing methods subjective objective evaluations.