Subset selection analysis for the reduction of hyperspectral imagery

Key Points

• The aim is to address the dimension reduction in hyperspectral imagery using subset selection as a matrix approximation problem.
• Formulated the dimension reduction problem as a matrix approximation challenge.
• Developed a heuristic algorithm to solve the subset selection problem.
• Utilized LANDSAT and AVIRIS images for numerical analysis.
• The selected bands are almost aligned with the first few principal components of the imagery.
• Demonstrated effective dimension reduction with significant computational efficiency.
• Numerical results validate the proposed heuristic algorithm's effectiveness.