Compressive adaptive ghost imaging via sharing mechanism and fellow relationship

Area of Science:

• Computational imaging
• Signal processing
• Wavelet analysis

Background:

• Ghost imaging (GI) is an advanced imaging technique.
• Traditional GI methods often require high sampling rates, limiting practical applications.
• Improving imaging quality while reducing sampling rates remains a key challenge.

Purpose of the Study:

• To propose a novel compressive adaptive ghost imaging (CAGI) method.
• To leverage wavelet tree properties for efficient sampling.
• To enhance ghost image quality and reduce sampling rates.

Main Methods:

• Utilizing intrascale and interscale sharing mechanisms and fellow relationships within a wavelet tree.
• Sampling shared coefficients based on parent coefficients and interscale relationships.
• Predicting and sampling significant coefficients using intrascale mechanisms.
• Reconstructing ghost images using significant coefficients and coarse images.

Main Results:

• Demonstrated improved imaging quality at equivalent sampling rates.
• Achieved lower sampling rates for comparable imaging quality.
• Validated performance across various object images in both noise-free and noisy conditions.
• Showcased the effectiveness of shared coefficients and predictive sampling.

Conclusions:

• The proposed CAGI method effectively reduces sampling rates.
• The method enhances ghost image quality by incorporating detailed information.
• Wavelet tree-based sampling strategies offer significant advantages for ghost imaging.