Motion-compensated single-pixel imaging based on moment-encoded illumination under low sampling rates

Area of Science:

• Optics and Photonics
• Image Processing
• Computational Imaging

Background:

• Motion blur significantly degrades image quality in single-pixel imaging (SPI), especially for high-speed targets.
• Low sampling rates in SPI exacerbate motion blur challenges, limiting applications in dynamic scenarios.

Purpose of the Study:

• To propose and validate a motion blur restoration method for SPI systems targeting high-speed moving objects.
• To enhance image reconstruction quality and motion parameter estimation accuracy under blur conditions.

Main Methods:

• A novel restoration strategy based on geometric moment patterns for SPI was developed.
• Real-time localization and geometric moment detection were employed to extract motion parameters.
• A pattern-shifting reconstruction algorithm was utilized to synchronize target trajectory with illumination patterns.

Main Results:

• The proposed method accurately estimates motion parameters and effectively suppresses motion blur.
• Significant improvements in velocity detection accuracy and reconstructed image quality were observed.
• The algorithm demonstrated robust performance across various motion conditions in simulations and experiments.

Conclusions:

• The geometric moment pattern-based method offers a feasible solution for high-speed SPI applications with motion blur.
• This approach enhances the practical utility of SPI for dynamic target imaging.
• The study validates the effectiveness of geometric moments in motion blur mitigation for low-sampling-rate imaging.