Generalized image deconvolution by exploiting the transmission matrix of an optical imaging system

Area of Science:

• Optical physics
• Image processing
• Computational imaging

Background:

• Optical imaging systems suffer from aberrations and defects, degrading image quality.
• Image deconvolution is a common recovery technique, often assuming linear shift-invariance.
• The shift-invariance assumption is not always valid and limits deconvolution performance.

Purpose of the Study:

• To develop a generalized method for image deconvolution that does not require shift-invariance.
• To address the limitations of current deconvolution techniques by fully characterizing optical systems.
• To enable distortion-free imaging under general conditions.

Main Methods:

• Proposed a generalized method to solve the linear inverse problem of coherent light propagation.
• Fully measured the transmission matrix of the imaging system.
• Avoided regularization methods and constraints on shift invariance.

Main Results:

• Successfully corrected severe aberrations caused by a tilted lens and a disordered layer.
• Demonstrated the effectiveness of the generalized deconvolution method without shift-invariance assumptions.
• Achieved distortion-free imaging under general imaging conditions.

Conclusions:

• The proposed method generalizes image deconvolution theory.
• Full characterization of the optical imaging system enables aberration compensation.
• This approach allows for high-quality imaging even with significant optical imperfections.