Ghost imaging under low-rank constraint.

Area of Science:

• Computational imaging
• Image processing
• Optical imaging

Background:

• Ghost imaging (GI) is an optical imaging technique that reconstructs images from intensity correlations.
• Compressive sensing (CS) is often used with GI, typically employing sparsity constraints for image reconstruction.
• Existing methods may face limitations in image quality, particularly with limited data.

Purpose of the Study:

• To introduce and evaluate a novel low-rank constraint for ghost imaging with compressive sensing (GICC).
• To compare the performance of low-rank constrained GICC against traditional sparsity constrained GICC (GISC).
• To assess the effectiveness of the low-rank constraint as a reference-less image enhancement tool.

Main Methods:

• A low-rank constraint was applied to the two-dimensional image reconstruction process in ghost imaging.
• Simulations and experimental setups were used to validate the proposed method.
• Image quality was assessed using metrics such as equivalent numbers of looks, focusing on smoothness and overall fidelity.
• The robustness of the low-rank parameter was investigated.

Main Results:

• The low-rank constraint demonstrated explicit improvement in ghost imaging quality compared to sparsity constraints, particularly in under-sampling situations.
• Enhanced image smoothness was observed, as quantified by the equivalent numbers of looks.
• The low-rank constraint proved robust across various parameter settings.
• The method functioned effectively as a reference-less image quality enhancement tool for GISC-reconstructed images.

Conclusions:

• The low-rank constraint offers a significant advantage over sparsity constraints for ghost imaging with compressive sensing.
• This approach yields superior image quality, especially under challenging under-sampling conditions.
• Low-rank constraint is a valuable technique for improving ghost imaging reconstruction and serves as an effective post-processing enhancement tool.