Summary
- Boston University researchers introduce a novel neural phase retrieval (NeuPh) approach using deep learning techniques.
- NeuPh reconstructs high-resolution phase information from low-resolution measurements using a CNN-based encoder and an MLP-based decoder.
- The method outperforms traditional model-based methods and current neural networks, providing robust resolution enhancement and accurate results.
- NeuPh exhibits strong generalization capabilities, showcasing adaptability to different experimental conditions.