Deep Learning Phase Aberration Compensation in Digital Holography
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital holography methods face challenges in accurately compensating phase aberrations before phase unwrapping, leading to dense fringes, fringe dislocations, and coherent noise, which reduce the reliability of phase data and introduce errors in the measurement of microstructural surfaces.
Innovation Solution
A digital holographic wrapped phase aberration compensation method based on deep learning, where a neural network model is trained with simulated wrapped phase maps and Zernike polynomial coefficients to automatically compensate most aberration components before phase unwrapping, eliminating the need for manual intervention and initial parameter input.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional phase aberration numerical compensation methods are used after phase unwrapping, then phase aberration can be compensated, but inaccurate phase unwrapping due to dense fringes and fringe dislocations restricts the effectiveness of compensation
Solution Approach 1:
The patent applies preliminary action by performing phase aberration compensation before phase unwrapping. The deep learning model directly processes the wrapped phase map to estimate Zernike polynomial coefficients, which are then used to compensate phase aberrations. This preliminary compensation removes dense fringes and reduces fringe dislocations before the unwrapping process, ensuring more reliable phase unwrapping and improving overall measurement precision.
2Reliability
If phase filtering is applied to denoise and smooth wrapped phase map before phase unwrapping, then noise is reduced, but inappropriate filtering may cause fringe damage, aggravate fringe misalignment or over-smooth the edge of sample structure
Solution Approach 1:
The patent replaces traditional mechanical phase filtering operations with a deep learning-based phase aberration compensation system. Instead of applying filters that may damage fringes or over-smooth edges, the neural network directly estimates phase aberrations from the wrapped phase map and compensates them, preserving sample structure edges while removing noise and reducing fringe dislocations.
3Productivity
If deep learning network model directly processes wrapped phase map to output Zernike polynomial coefficients, then fast calculation and high accuracy are achieved, but complex network training and model establishment are required
Solution Approach 1:
The patent changes the parameter representation by directly estimating Zernike polynomial coefficients from the wrapped phase map using a deep learning model. This parameter transformation enables fast calculation during measurement since the trained model can quickly process new samples. The complexity of network training is a one-time cost that enables rapid, accurate phase aberration compensation for subsequent measurements.
Data Source
AI summary
In a digital holographic wrapped phase aberration compensation method based on deep learning, a random Zernike polynomial coefficient and a corresponding wrapped phase map are generated by a computer and are respectively treated as a learning label and a network to train a neural network model. A digital holographic optical setup is built to record a hologram of a sample to be measured, the wrapped phase map is inputted into the trained neural network model after numerical reconstruction, and the Zernike polynomial coefficient is outputted to reconstruct a phase aberration distribution and to compensate complex amplitude in a spatial domain. Phase filtering and unwrapping are performed on the compensated wrapped phase map, and Zernike polynomial fitting based on background segmentation is performed on the unwrapped phase to compensate for residual aberration.