Collaborative-guided spectral abundance learning with bilinear mixing model for hyperspectral subpixel target...

Area of Science:

• Remote Sensing
• Hyperspectral Imaging
• Signal Processing

Background:

• Subpixel target detection in hyperspectral images is challenging.
• Linear mixing models fail to capture nonlinear material interactions.
• Existing methods often yield unsatisfactory performance due to model limitations.

Purpose of the Study:

• To develop a novel model for accurate subpixel target detection.
• To address the limitations of linear mixing models in hyperspectral data.
• To improve the performance of hyperspectral subpixel target detection.

Main Methods:

• Proposed a collaborative-guided spectral abundance learning model (CGSAL).
• Utilized a bilinear mixing model to account for nonlinear scattering.
• Incorporated a collaborative term to enhance spectral abundance learning.

Main Results:

• CGSAL effectively models nonlinear scattering and material interactions.
• The model achieved accurate spectral abundance learning.
• Experiments on real-world and synthetic datasets demonstrated competitive performance.

Conclusions:

• CGSAL offers a significant advancement in subpixel target detection.
• The proposed method outperforms state-of-the-art hyperspectral subpixel target detectors.
• CGSAL provides a robust solution for complex hyperspectral imaging scenarios.