EigenCWD: a spatially varying deconvolution algorithm for single metalens imaging

Area of Science:

• Optics and Photonics
• Computational Imaging
• Image Processing

Background:

• Two-dimensional metalenses offer miniaturization in optics, enabling new imaging applications.
• Single-lens imaging is prevalent, but metalens limitations cause aberrations and require computational deconvolution.
• Spatially varying aberrations like coma and astigmatism in metalens imaging are challenging for standard deconvolution.

Purpose of the Study:

• To develop an advanced deconvolution algorithm for correcting spatially varying aberrations in metalens imaging.
• To overcome the limitations of traditional deconvolution methods like Wiener filtering for metalens-generated distortions.

Main Methods:

• Developed a spatially varying deconvolution algorithm named eigenvalue column-wise decomposition (eigenCWD).
• Utilized an approximate forward blurring model with eigendecomposition of spatially varying point spread functions for efficient computation.
• Applied eigenCWD to solve image reconstruction minimization problems.

Main Results:

• EigenCWD effectively corrects spatially varying blur and distortions in images from metalenses.
• The algorithm demonstrates superior performance compared to the Wiener filter for complex aberrations.
• Efficient computation allows scaling to large image sizes and blurring kernels.

Conclusions:

• EigenCWD offers a robust solution for deblurring and distortion correction in metalens imaging.
• This method enhances image quality by addressing limitations of current deconvolution techniques.
• The algorithm facilitates broader adoption of metalenses in advanced imaging applications.