Denoised single-pixel imaging in a Fourier acquisition mode

Area of Science:

• Optics
• Image Processing
• Computational Imaging

Background:

• Environmental noise significantly degrades imaging quality in Fourier Single Pixel Imaging (FSPI).
• Existing denoising methods may not sufficiently address noise interference in FSPI reconstruction.
• Robust and efficient imaging schemes are needed to mitigate noise impacts.

Purpose of the Study:

• To propose and evaluate simple, efficient denoised single-pixel imaging schemes using linear filters.
• To assess the effectiveness of average and Gaussian filters in reducing environmental noise in FSPI.
• To compare the performance of the proposed schemes against conventional and deringing SPI methods.

Main Methods:

• Implementation of two denoised FSPI schemes: SCH-A (average filter) and SCH-G (Gaussian filter).
• Experimental and simulation-based analysis of image reconstruction quality under noisy conditions.
• Evaluation of filtering effects with varying template sizes and variances.

Main Results:

• Both SCH-A and SCH-G effectively reduce environmental noise impact, showing greater robustness than conventional and deringing SPI.
• SCH-G preserves more image details and edges compared to SCH-A.
• Increasing template size enhances filtering effects for both schemes under consistent variance.

Conclusions:

• Linear filtering, specifically using average and Gaussian filters, offers a simple and effective approach to denoise FSPI.
• SCH-G provides superior detail retention over SCH-A, making it preferable for applications requiring high fidelity.
• Template size is a critical parameter for optimizing filter performance in denoised FSPI schemes.