Neural-Network Based Reflector Design for Finite 2D Sources in Far-Field Illumination
A novel computational approach for designing reflectors for finite two-dimensional light sources in far-field illumination is proposed. The reflector is parameterized as a multilayer perceptron, and an energy-conserving loss function is optimized, outperforming deconvolution methods.
Authors
- Martijn Anthonissen (ORCID: https://orcid.org/0000-0003-3667-769X)
- Koondanibha Mitra (ORCID: https://orcid.org/0000-0002-8264-5982)
- Lisa Kusch (ORCID: https://orcid.org/0000-0001-7558-0618)
- Wilbert L. IJzerman
- Roel Hacking (ORCID: https://orcid.org/0009-0001-0820-2614)
Institutions
- Signify (Netherlands) (NL)
- Eindhoven University of Technology (NL)
Publication Details
- Published
- 2025-01-01
- DOI
- https://doi.org/10.1364/freeform.2025.frw4a.4
- Primary Topic
- Optical measurement and interference techniques
- Type
- article
- Field-Weighted Citation Impact
- 0.00