Pansharpening Applications in Ecological and Environmental Monitoring Using an Attention Mechanism-Based Dual-Stream Cross-Modality Fusion Network

Pansharpening is a critical technique in remote sensing, particularly in ecological and environmental monitoring, where it is used to integrate panchromatic (PAN) and multispectral (MS) images. This technique plays a vital role in assessing environmental changes, monitoring biodiversity, and supporting conservation efforts. While many current pansharpening methods primarily rely on PAN images, they often overlook the distinct characteristics of MS images and the cross-modal relationships between them. To address this limitation, the paper presents a Dual-Stream Cross-modality Fusion Network (DCMFN), designed to offer reliable data support for environmental impact assessment, ecological monitoring, and material optimization in nanotechnology. The proposed network utilizes an attention mechanism to extract features from both PAN and MS images individually. Additionally, a Cross-Modality Feature Fusion Module (CMFFM) is introduced to capture the complex interrelationships between PAN and MS images, enhancing the reconstruction quality of pansharpened images. This method not only boosts the spatial resolution but also maintains the richness of multispectral information. Through extensive experiments, the DCMFN demonstrates superior performance over existing methods on three remote sensing datasets, excelling in both objective evaluation metrics and visual quality.