Optimization of retina-like illumination patterns in ghost imaging.

Area of Science:

• Optics
• Image Reconstruction
• Computational Imaging

Background:

• Ghost imaging (GI) offers advantages like scattering robustness and wide spectrum imaging.
• Traditional GI requires numerous measurements for high-resolution images.
• Compressive sensing approaches, like retina-like patterns, aim to reduce measurement requirements.

Purpose of the Study:

• To enhance the imaging quality of the region of interest (ROI) in ghost imaging.
• To develop an optimized retina-like pattern design for improved GI performance.
• To reduce the number of measurements needed for sharp GI reconstruction.

Main Methods:

• Proposed a novel method for optimizing retina-like patterns by incorporating object sparsity priors into the ROI.
• Compared the proposed method with conventional GI, standard retina-like GI, and GI using principal component analysis (PCA)-optimized patterns.
• Verified the method through simulations and experimental validations.

Main Results:

• The optimized retina-like pattern method achieved superior imaging quality in the ROI compared to all other tested methods.
• Demonstrated good generalization capability of the optimized retina-like patterns.
• Showcased the ability to extract feature information by optimizing ROI size and position.

Conclusions:

• Optimized retina-like patterns, incorporating object sparsity, are highly effective for high-quality ghost imaging.
• This approach significantly improves ROI image reconstruction while maintaining measurement efficiency.
• The method facilitates practical realization of high-quality ghost imaging applications.