Abstract:Pansharpening belongs to an important part in the field of remote sensing image fusion, which refers to the fusion of low spatial resolution multispectral (MS) images with high spatial resolution (PAN) images to finally obtain high resolution multispectral (HRMS) images. The existing deep learning-based pan-sharpening methods have achieved better results compared with the traditional methods, but there are still two problems: spectral distortion and loss of spatial detail information. We propose an end-to-end attention-based dual-residual multi-stage remote sensing image fusion network (ADRPN) for MS image and PAN image pan-sharpening based on an in-depth study of spectral information of MS images and spatial information fusion of PAN images, which consists of three main stages, each resembling an encoder-decoder, and cross-stage fusion to achieve channel domain feature stitching. The first two stages are the feature extraction stage, where the features of MS images and PAN images are extracted using the residual module, and the different features learned are used to guide the training of individual networks using CRFB (Cross residual feature block). In the third stage (image reconstruction), we use the CA (Coordinate attention) and SE (Squeeze-and-Excitation) attention mechanism to enable the network to more precisely locate the region of interest by the precise location information obtained, which allows the features extracted in the first two stages to be better fused with the original image, thus reducing the occurrence of spectral distortion and loss of spatial detail information. Qualitative and quantitative analyses of real and simulated data from the benchmark datasets QuickBird (QB), GF-2, and WorldView-2 (WV2) show that the method can better preserve the spectral and spatial detail information and obtain high-quality HRMS images.

MC-JAFN: Multilevel Contexts-Based Joint Attentive Fusion Network for Pansharpening

An Efficient Pansharpening Approach Based on Spatial Information and Multi-Layer Fusion.

Pansharpening Based on Multi-Branch CNN

MESSFN : a Multi-level and Enhanced Spectral-Spatial Fusion Network for Pan-sharpening

Multi-Scale Bilateral Attention Fusion Network for Pansharpening

Pan-Sharpening Based on Multilevel Coupled Deep Network

Multispectral Pan-sharpening via Dual-Channel Convolutional Network with Convolutional LSTM Based Hierarchical Spatial-Spectral Feature Fusion

Attention-based dual residual network-based for multi-spectral pan-sharpening

Multi-level features fusion via cross-layer guided attention for hyperspectral pansharpening

A Multiscale and Multidepth Convolutional Neural Network for Remote Sensing Imagery Pan-Sharpening.

Going Deeper with Densely Connected Convolutional Neural Networks for Multispectral Pansharpening

PMACNet: Parallel Multiscale Attention Constraint Network for Pan-Sharpening

A Multi-Scale Densely Deep Learning Method for Pansharpening.

Coarse-to-Fine Cross-Domain Learning Fusion Network for Pansharpening.

Attention_FPNet: Two-Branch Remote Sensing Image Pansharpening Network Based on Attention Feature Fusion.

DCPNet: A Dual-Task Collaborative Promotion Network for Pansharpening

Asymmetric Bidirectional Fusion Network for Remote Sensing Pansharpening

A Two-Stream Multiscale Deep Learning Architecture for Pan-Sharpening

MPEFNet: Multilevel Progressive Enhancement Fusion Network for Pansharpening

Convolutional LSTM-Based Hierarchical Feature Fusion for Multispectral Pan-Sharpening

Multibranch Cnn-Based Pansharpening with Skip Connection