Image-free tracking of single-pixel detection in dynamic complex scenes.

Area of Science:

• Optics and Photonics
• Image Processing
• Computational Imaging

Background:

• Single-pixel imaging (SPI) faces performance degradation with complex backgrounds.
• Existing methods for dynamic target tracking often require impractical background pre-acquisition and subtraction.
• Robust tracking against unknown, dynamic backgrounds remains a significant challenge in SPI.

Purpose of the Study:

• To develop a Fourier single-pixel imaging method for robustly tracking moving targets against dynamic complex backgrounds.
• To improve tracking precision and trajectory stability without prior background knowledge.
• To overcome the limitations of existing methods that require background pre-acquisition.

Main Methods:

• Fourier single-pixel imaging combined with intermediate frequency filtering.
• Application of a band-pass filter to sparsely sampled Fourier data to suppress low-frequency background interference.
• One-dimensional optical flow algorithm designed to estimate target displacement from filtered projections.

Main Results:

• Effective suppression of low-frequency background interference.
• Robust estimation of target displacement without image reconstruction.
• Significant improvement in tracking robustness, precision, and trajectory stability.
• Successful imaging and tracking of moving targets against complex, low-rank backgrounds in simulations and experiments.

Conclusions:

• The proposed Fourier SPI method offers a practical solution for tracking moving targets in challenging background conditions.
• The integration of intermediate frequency filtering and 1D optical flow enhances tracking performance.
• This approach advances the capabilities of single-pixel imaging for real-world dynamic scene analysis.