Towards self-calibrated lens metrology by differentiable refractive deflectometry

Area of Science:

• Optical Metrology
• Computational Imaging
• Lens Design and Manufacturing

Background:

• Deflectometry is a non-contact optical technique for surface measurement.
• Measuring refractive elements with deflectometry is challenging due to multi-surface interactions and strict pose requirements.
• Existing methods struggle with the complexity of refractive optics.

Purpose of the Study:

• To develop a computational self-calibration approach for measuring parametric lenses.
• To enable accurate measurement of lens shape and pose using dual-camera refractive deflectometry.
• To overcome the limitations of traditional deflectometry for refractive optical elements.

Main Methods:

• Utilized a dual-camera refractive deflectometry setup.
• Developed an accurate, differentiable, and efficient ray tracing framework to model the metrology system.
• Employed damped least squares for estimating unknown lens shape and pose parameters.

Main Results:

• Successfully demonstrated the computational self-calibration approach on singlet lenses.
• Validated the method for measuring surface curvature and asphere-freeform optics in a transmissive setting.
• Achieved accurate estimation of lens shape and pose parameters through synthetic and experimental data.

Conclusions:

• The proposed computational self-calibration method effectively measures refractive lenses.
• This technique advances optical metrology for complex lens systems.
• The approach provides a robust solution for characterizing lens shape and pose.