Enhancing remote sensing image fusion with deep learning attention fusion residual approach

Abstract

Abstract This research introduces an innovative deep learning framework aimed at improving remote sensing image fusion through the Attention Fusion Residual (AFR) technique. The model employs attention mechanisms to prioritize critical features from the source images, while the residual network facilitates the preservation of intricate details. By merging multi-spectral and panchromatic images, the AFR technique enhances both spatial and spectral quality. Experimental findings indicate that the proposed method surpasses conventional and leading-edge techniques in terms of visual quality and quantitative metrics such as Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM), representing a noteworthy progression in the specific field of image processing and remote sensing. Thus the fused images demonstrate enhanced detail and better noise suppression compared to traditional fusion techniques. These improvements are particularly evident in the enhanced visibility of fine details and in maintaining the color fidelity of the landscape, which are critical for applications in environmental monitoring and urban planning.