SSCDN: a spatial-spectral collaborative network for hyperspectral image denoising.

Area of Science:

• Remote Sensing
• Medical Diagnostics
• Biological Research

Background:

• Hyperspectral imaging (HSI) captures full spectral data, valuable in various fields.
• Image acquisition noise degrades HSI quality, hindering analysis.
• Existing denoising methods struggle with complex noise and spectral recovery.

Purpose of the Study:

• To develop an advanced hyperspectral image denoising method.
• To enhance spatial and spectral feature preservation in degraded HSIs.
• To improve robustness against complex noise in HSI datasets.

Main Methods:

• Proposed a spatial-spectral collaborative denoising network (SSCDN).
• Integrated attention mechanisms and a novel collaborative attention module.
• Employed a multi-loss joint optimization strategy.

Main Results:

• SSCDN demonstrated superior denoising performance on simulated and real HSI data.
• The method effectively preserved spectral and spatial features.
• Achieved high structural-spectral fidelity and robustness across diverse noise levels.

Conclusions:

• The SSCDN model significantly outperforms state-of-the-art HSI denoising techniques.
• It offers an effective solution for complex noise challenges in hyperspectral imaging.
• The approach shows promise for remote sensing and biomedical applications.