MDCwFB: A Multilevel Dense Connection Network with Feedback Connections for Pansharpening

In most practical applications of remote sensing images, high-resolution multispectral images are needed. Pansharpening aims to generate (MS) from the input high spatial resolution single-band panchromatic (PAN) and low images. Inspired by remarkable results other researchers in pansharpening based on deep learning, we propose a multilevel dense connection network with feedback connection. Our consists four parts. The first part two identical subnetworks extract features PAN MS second is feature fusion recovery network, which used fuse domain encode decode at different levels so that can fully capture information. third continuous operation, refines shallow feedback. fourth an image reconstruction network. High-quality recovered making full use multistage decoding through connections. Experiments satellite datasets show our proposed method superior existing methods, subjective visual evaluation objective indicators. Compared models, achieve significant gains multiple index values measure spectral quality details generated image, namely angle mapper (SAM), relative global dimensional synthesis error (ERGAS), structural similarity (SSIM).