Complex Amplitude Reading From Single Diffraction Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current detectors can only measure light intensity and require complex optical systems or iterative methods to obtain phase information, leading to unstable results, low accuracy, and slow computation speeds, which are inadequate for real-time image processing and computational imaging.
Innovation Solution
A non-interferometric, non-iterative method using a single diffraction pattern and a neural network model to directly extract complex amplitude information from intensity images, eliminating the need for reference beams and iterative processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If interference methods are used to obtain phase information, then phase information can be obtained, but the optical system becomes complex and the results become unstable
Solution Approach 1:
The patent extracts phase information directly from intensity images through neural network processing, removing the need for complex interference optical systems. The neural network model learns the mapping relationship between intensity patterns and phase information, enabling direct extraction without reference beams or interference patterns.
Solution Approach 2:
The patent replaces the mechanical/optical interference system with a computational neural network system. Instead of using physical interference patterns to encode phase information, the system uses trained neural networks to decode phase information directly from intensity measurements, substituting optical complexity with computational processing.
2Device complexity
If non-interference iterative methods are used to obtain phase information, then the optical system is simplified, but the calculation speed becomes slow
Solution Approach 1:
The patent performs preliminary training of the neural network model using iterative methods during the offline phase. Once trained, the model contains pre-learned phase retrieval knowledge that enables rapid direct inference during online operation, eliminating the need for iterative calculations during actual measurement and achieving both simplicity and speed.
Solution Approach 2:
The patent creates a computational copy of the phase retrieval process through the neural network model. The network learns to replicate the function of iterative phase retrieval algorithms but executes the copied function much faster during inference, trading offline training time for online processing speed.
3Device complexity
If non-interference iterative methods are used to obtain phase information, then the optical system is simplified, but the accuracy becomes low
Solution Approach 1:
The patent incorporates feedback mechanisms during the neural network training phase, where the network continuously adjusts its parameters based on the error between predicted and actual phase information. This feedback-driven training process enables the model to learn accurate phase retrieval mappings, achieving high precision without requiring complex optical feedback systems during measurement.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This method simplifies the optical system, enhances accuracy and computation speed, and enables simultaneous and stable reading of amplitude and phase information, suitable for applications in holographic storage and biomedical imaging.
Implementation Method 1
Step S01: Diffracting a light beam containing amplitude information and phase information to obtain a diffraction pattern with intensity variations
Data Source
AI summary
The present invention discloses a non-interferometric, non-iterative complex amplitude reading method and apparatus. The reading method includes the following steps: diffracting a light beam containing amplitude information and phase information to obtain a diffraction pattern with intensity variations; constructing a diffraction intensity-complex amplitude model and training it based on the correlation between the diffraction pattern and amplitude information and phase information, and applying the trained model directly to new diffraction images to obtain amplitude information and phase information. The method can achieve detection of complex amplitude information, including amplitude and phase, from a single diffraction image, improve the stability and accuracy of phase reading results, increase the calculation speed, and simplify the optical system. It is suitable for applications in holographic storage, biomedical image processing, and microscopic imaging, among others.