Pan-sharpening of Multispectral Remote Sensing Imagery Using Deep Back-Projection Network

摘要

This paper presents the pan-sharpening method using deep back-projection-based network. In the proposed method, the deep back-projection network plays a role of enlarging the spatial resolution of low-resolution multispectral band image preserving the spectral information. Next, we aggregate the high-resolution features of the panchromatic band image estimated by the convolutional layers. Experimental results show that the proposed method can provide improved pan-sharpened image preserving the details of panchromatic band image without spectral distortion.