Denoising in SVD-based ghost imaging.

Area of Science:

• Optics and photonics
• Image processing
• Computational imaging

Background:

• Ghost imaging (GI) efficiently reconstructs images using singular-valued decomposition (SVD).
• Noise significantly degrades image quality in GI, potentially obscuring object information.
• Existing SVD methods are sensitive to noise, limiting practical applications.

Purpose of the Study:

• To investigate truncated SVD (TSVD) for enhancing image quality in GI.
• To analyze noise effects on TSVD-based GI.
• To introduce filtering techniques to further improve TSVD-based GI performance.

Main Methods:

• Experimental implementation of TSVD for GI by selecting dominant singular values.
• Systematic analysis of noise impact on TSVD reconstruction.
• Application of additional filters to mitigate noise in TSVD-based GI.

Main Results:

• TSVD significantly improves image quality, enhancing contrast-to-noise ratio and structural similarity.
• Optimal truncation ratios were identified for effective noise suppression.
• Noise analysis revealed key factors affecting TSVD performance.
• Additional filters further boosted image quality in TSVD-based GI.

Conclusions:

• TSVD is an effective technique for improving ghost imaging quality under noisy conditions.
• The developed method offers a robust approach for rapid imaging in complex environments.
• TSVD-based GI shows promise for applications requiring high-quality imaging with reduced noise.