Hybrid diffractive optics design via hardware-in-the-loop methodology for achromatic extended-depth-of-field imaging.

Area of Science:

• Computational imaging and optics design
• Digital optics and computational photography

Background:

• Diffractive optical elements (DOEs) offer compact optical setups but face limitations from manufacturing errors and simulation mismatches.
• Existing digital differentiable models for DOE optimization are computationally intensive, restricting the designable DOE size.

Purpose of the Study:

• To develop a co-designed hybrid optical system and image reconstruction algorithm for improved imaging performance.
• To overcome limitations of current DOE design methods by integrating hardware-in-the-loop optimization.

Main Methods:

• A hardware-in-the-loop strategy was employed, optimizing a phase-only spatial light modulator (SLM) as a DOE with a refractive lens.
• A convolutional neural network with quantitative and qualitative loss functions was used for optimization.
• Physical light propagation was utilized instead of numerical modeling to bridge simulation-experimental gaps.

Main Results:

• The proposed system achieved advanced all-in-focus sharp imaging over a depth range of 0.4-1.9 meters.
• Performance was validated against multi-lens optics in commercial smartphones and mirrorless cameras.
• The hardware-in-the-loop approach effectively mitigated issues related to manufacturing artifacts and calibration errors.

Conclusions:

• Co-design of hybrid optics and reconstruction algorithms using hardware-in-the-loop optimization offers a robust solution for compact imaging systems.
• This method enhances DOE design by integrating physical feedback, overcoming limitations of purely digital approaches.
• The system demonstrates superior performance in achieving depth-agnostic sharp imaging compared to conventional optics.