Learned digital lens enabled single optics achromatic imaging.

Area of Science:

• Optics and Photonics
• Computational Imaging
• Computer Vision

Background:

• Advanced imaging systems often face performance limitations due to optical complexity and aberrations.
• Existing methods struggle to balance optical simplicity with high-fidelity image reconstruction.

Purpose of the Study:

• To develop a novel digital lens (DL) framework for compensating optical aberrations.
• To enable achromatic imaging with reduced optical complexity using computational methods.

Main Methods:

• Proposed a digital lens (DL) to correct chromatic aberration computationally.
• Integrated a self-designed neural network to address residual wavelength-independent degradations.
• Transformed physical aberration correction into an algorithm-based computational imaging task.

Main Results:

• Achieved achromatic imaging in the analog domain with a simplified optical setup.
• Demonstrated up to 14.62 dB higher peak signal-to-noise ratio (PSNR) compared to original chromatic images.
• Showcased a 4.83 dB PSNR improvement by incorporating the digital lens compared to methods without it.

Conclusions:

• The proposed digital lens framework offers a viable alternative for applications demanding simple optics and high-fidelity visualization.
• This computational imaging approach effectively reduces optical complexity while enhancing image quality.
• The method paves the way for future advancements in academic and industrial imaging research.