Deep neural network based calibration for freeform surface misalignments in general interferometer.

Area of Science:

• Optical Engineering
• Metrology
• Surface Metrology

Background:

• Interferometric testing of optical freeform surfaces presents significant challenges in calibrating misalignment aberrations.
• Complex surface figures and six-axis degrees of freedom lead to non-linear aberration variations with minor misalignments.
• Traditional methods like the sensitive matrix approach lack precision in non-linear regimes.

Purpose of the Study:

• To introduce a deep neural network (DNN)-based method for accurate calibration of surface misalignment aberrations in optical freeform testing.
• To address the limitations of traditional methods in handling non-linear aberration variations.
• To demonstrate the feasibility and accuracy of the proposed DNN-based calibration technique.

Main Methods:

• A deep neural network (DNN) was trained to learn the non-linear relationship between surface misalignments and aberrations.
• Ray tracing simulations were used to predict misalignment aberrations based on estimated misalignments.
• Wavefront data subtraction was employed to remove the predicted aberrations.

Main Results:

• The DNN-based method accurately estimates non-linear misalignment aberrations, surpassing traditional techniques.
• Simulated and experimental results confirm the effectiveness of the DNN approach.
• The method allows for precise calibration and correction of aberrations caused by surface misalignments.

Conclusions:

• The proposed DNN-based calibration method is a feasible and accurate solution for addressing misalignment aberrations in optical freeform surface testing.
• This technique offers a significant improvement over traditional methods, particularly for complex, non-linear scenarios.
• The study highlights the potential of machine learning in advancing optical metrology.