Image quality of compressive single-pixel imaging using different Hadamard orderings

Area of Science:

• Optics and Photonics
• Computational Imaging
• Signal Processing

Background:

• Single-pixel imaging (SPI) is an emerging technique gaining traction across various scientific fields.
• SPI systems capture scenes using a single-pixel detector, relying on computational reconstruction.
• The choice of basis functions and their ordering significantly impacts reconstruction fidelity.

Purpose of the Study:

• To evaluate the influence of different Hadamard basis orderings on single-pixel imaging reconstruction quality.
• To compare the performance of TVAL3 and NESTA reconstruction algorithms under varying conditions.
• To assess the impact of noise levels and compression ratios on image reconstruction.

Main Methods:

• Utilized simulation and experimental single-pixel camera setups employing a digital micromirror device.
• Tested five Hadamard basis orderings: Natural, Walsh, Cake-cutting, High Frequency, and Random.
• Evaluated two reconstruction algorithms (TVAL3, NESTA) across three noise levels and compression ratios (0.1 to 1).

Main Results:

• For a compression ratio of 0.1, Cake-cutting order yielded the best reconstruction quality, while Walsh order provided the best contrast.
• Walsh and Cake-cutting orders demonstrated comparable results at a compression ratio of 0.5.
• Good image quality was achieved with Walsh and Cake-cutting orders for compression ratios as low as 0.3.
• The TVAL3 algorithm consistently outperformed NESTA in image reconstruction quality for compression ratios between 0.1 and 0.5.

Conclusions:

• Hadamard basis ordering critically affects single-pixel imaging performance.
• Walsh and Cake-cutting orders are highly effective for SPI, particularly at reduced sampling rates.
• TVAL3 is a more robust reconstruction algorithm than NESTA for SPI applications within the tested compression ranges.