High-precision large-aperture single-frame interferometric surface profile measurement method based on deep learning

Highlights This is the first time that high-precision measurement of large-aperture optical components has been realized by capturing single-frame interferograms without requiring a phase shifter. We integrate deep learning algorithms with interferometry methods to simulate the interferometric process in a deep learning framework. This approach mitigates environmental noise effects on measurement accuracy, eliminates phase shifters, and enables dynamic surface profile measurements of large-aperture optical components. Compared with traditional phase-shifting methods, this approach achieves a 48-fold time reduction while improving measurement efficiency and stability. We demonstrate a highly efficient dynamic measurement method that achieves comparable accuracy without ZYGO interferometers’ ultra-stable environment requirements.