Scaling-Based Two-Step Reconstruction in Full Polarization-Compressed Hyperspectral Imaging

Area of Science:

• Optical Engineering
• Signal Processing
• Hyperspectral Imaging

Background:

• Polarized hyperspectral imaging captures rich target physicochemical characteristics but poses signal processing challenges.
• Existing compressive sensing methods struggle to reconstruct complete polarization information due to system simplifications.
• Accurate reconstruction of full polarization information is crucial for advanced optical sensing applications.

Discussion:

• A two-step reconstruction method is proposed to progressively handle polarization characteristics at different scales.
• The method utilizes a quarter-wave plate and liquid crystal tunable filter for full polarization compression and hyperspectral imaging.
• Stokes parameters and modulation coefficients are scaled based on numerical features for optimized reconstruction.

Key Insights:

• The first Stokes parameter is reconstructed using compressive sensing in the initial step.
• Subsequent Stokes parameters with similar magnitudes are reconstructed in the second step, leveraging prior results.
• This progressive approach addresses the challenge of reconstructing diverse polarization information.

Outlook:

• Simulation results demonstrate a 7.6 dB improvement in reconstruction accuracy for previously unachievable parameters.
• Reconstruction time is reduced by 8.25 hours without compromising high accuracy.
• The feature scaling method offers a valuable reference for fast, high-quality reconstruction of physical quantities with significant numerical disparities.