Noise-robust coded-illumination imaging with low computational complexity.

Area of Science:

• Optical Imaging
• Computational Imaging
• Image Reconstruction

Background:

• Coded-illumination (CI) imaging enhances resolution and information extraction in optical systems.
• Current CI reconstruction methods face challenges balancing noise robustness and computational efficiency.
• Practical CI applications require methods that are both noise-robust and computationally inexpensive.

Purpose of the Study:

• To propose a novel reconstruction scheme for coded-illumination imaging.
• To address the trade-off between noise robustness and low computational complexity in existing methods.
• To develop a scheme suitable for practical, high-dimensional optical imaging applications.

Main Methods:

• An iterative reconstruction scheme with two phases: weighted solving and alternating projection.
• A non-uniform and adaptive weighted solver to identify and attenuate measurement noise.
• Alternating projection optimization with support constraints, avoiding matrix lifting.

Main Results:

• The proposed scheme demonstrates strong noise robustness against various measurement noise types.
• Achieved low computational complexity and fast convergence in simulations and experiments.
• Successfully applied to structured illumination imaging and Fourier ptychography.

Conclusions:

• The novel CI reconstruction scheme offers significant improvements in noise handling and computational efficiency.
• The method provides a practical solution for high-dimensional information extraction in optical systems.
• The scheme is adaptable for diverse incoherent and coherent CI imaging modalities.