Correlation-informed ordered dictionary learning for imaging in complex media

Area of Science:

• Wave physics
• Computational imaging
• Signal processing

Background:

• Imaging in scattering media is challenging due to signal distortion.
• Existing methods often require prior knowledge of source locations or properties.
• Large and diverse datasets are crucial for advanced imaging techniques.

Purpose of the Study:

• To develop a robust imaging method for scattering media.
• To enable image reconstruction without prior knowledge of source parameters.
• To achieve high-resolution imaging in complex environments.

Main Methods:

• A two-step approach combining dictionary learning and multidimensional scaling.
• Estimating Green's function vectors using dictionary learning from unordered array data.
• Ordering sensing matrix columns via multidimensional scaling and cross-correlations for image reconstruction.

Main Results:

• Successfully estimated true Green's function vectors from sparse, unknown source data.
• Ordered sensing matrix columns to enable accurate imaging.
• Achieved image resolution comparable to homogeneous media in simulations.

Conclusions:

• The proposed method effectively images complex scattering media.
• It overcomes limitations of unknown source information.
• Demonstrates potential for high-resolution imaging in challenging environments.