Cross-Scale Spectral Calibration for Spatiotemporal Fusion of Remote Sensing Images.

Area of Science:

• Remote Sensing
• Image Processing
• Geospatial Analysis

Background:

• Spatiotemporal fusion integrates multi-source remote sensing data to achieve high spatial and temporal resolution.
• Existing methods often fail to address spectral inconsistencies between coarse- and fine-resolution images, degrading fusion results.
• Cross-scale spectral discrepancy is a significant challenge in spatiotemporal fusion.

Purpose of the Study:

• To propose a novel framework, XSC-Net, for spatiotemporal fusion that explicitly models and corrects cross-scale spectral discrepancies.
• To improve the radiometric fidelity and temporal reliability of fused remote sensing images.
• To enhance the performance of spatiotemporal fusion by addressing spectral inconsistencies.

Main Methods:

• Developed a cross-scale spectral calibration framework (XSC-Net).
• Introduced a spatial feature refinement block to enhance spatially discriminative structures.
• Incorporated a hierarchical spectral refinement block for adaptive channel-wise spectral calibration.

Main Results:

• XSC-Net effectively suppresses spectral distortion while preserving fine spatial details.
• The proposed method demonstrates superior performance compared to state-of-the-art spatiotemporal fusion techniques.
• Experiments on CIA and LGC datasets validate the effectiveness of XSC-Net.

Conclusions:

• XSC-Net provides an effective solution for addressing cross-scale spectral inconsistencies in spatiotemporal fusion.
• The framework significantly enhances the quality of fused remote sensing images.
• Ablation studies confirm the contribution of individual architectural components to the overall performance.