Noise-robust and data-efficient compressed ghost imaging via the preconditioned S-matrix method.

Area of Science:

• Optics
• Image Processing
• Computational Imaging

Background:

• Illumination pattern design is critical for ghost imaging (GI) quality.
• S-matrix binary patterns are suitable for GI with non-visible light due to modulator limitations.
• Limited research exists on compressed GI using S-matrix, particularly regarding sampling rate and noise resistance.

Purpose of the Study:

• To investigate the performance of S-matrix compressed ghost imaging (SCSGI).
• To propose a post-processing method, preconditioned S-matrix compressed GI (PSCSGI), to enhance SCSGI.
• To improve imaging quality and data efficiency of SCSGI, especially under noisy conditions.

Main Methods:

• Investigated SCSGI performance with S-matrix illumination patterns.
• Developed and applied a post-processing technique, PSCSGI.
• Conducted simulations and experiments to evaluate PSCSGI against SCSGI.

Main Results:

• PSCSGI improves imaging quality in noisy environments compared to SCSGI.
• PSCSGI achieves comparable or better results using only half the data of SCSGI.
• Effective image reconstruction is possible with PSCSGI even at a 5% sampling rate.

Conclusions:

• PSCSGI enhances imaging quality and data efficiency for S-matrix based compressed GI.
• The proposed method demonstrates robustness in noisy conditions and with low sampling rates.
• PSCSGI is expected to promote the use of S-matrix binary masks in ghost imaging applications.