Real-time phase-retrieval and wavefront sensing enabled by an artificial neural network

Area of Science:

• Optics and Photonics
• Computational Imaging
• Machine Learning Applications

Background:

• Wavefront reconstruction is crucial for optical system alignment and aberration correction.
• Traditional iterative phase retrieval methods are time-consuming, limiting real-time applications.
• Real-time wavefront sensing is essential for adaptive optics and dynamic optical adjustments.

Purpose of the Study:

• To demonstrate a real-time wavefront reconstruction method using a neural network.
• To overcome the speed limitations of conventional iterative phase retrieval techniques.
• To enable live adjustment and correction in complex optical systems.

Main Methods:

• A neural network was trained using simulated diffraction pattern data.
• The trained neural network was experimentally validated.
• Wavefronts were reconstructed from diffraction patterns behind a mask in real-time.

Main Results:

• The neural network achieved wavefront reconstruction in milliseconds, a significant speed improvement over iterative methods (seconds).
• The method demonstrated superior performance compared to iterative phase retrieval, especially with noisy diffraction patterns.
• Real-time reconstruction facilitates dynamic adjustments and corrections in optical setups.

Conclusions:

• Neural network-based phase retrieval offers a fast and robust solution for real-time wavefront reconstruction.
• This method enhances the capabilities of adaptive optics and complex optical system control.
• The approach shows promise for applications requiring rapid optical wavefront analysis and correction.