Light Correction Coefficient Prediction via Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for correcting optical aberrations are computationally intensive, leading to increased calculation time, which needs to be reduced for efficient aberration correction.
Innovation Solution
A light correction coefficient prediction method that involves acquiring intensity distributions from regions of interest in an intensity image, generating comparison data between actual and target distributions, and using a learning model to predict coefficients for aberration correction, thereby shortening calculation time and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods are used to derive coefficients for aberration correction, then accurate correction can be achieved, but the calculation time increases
Solution Approach 1:
The patent applies preliminary action by pre-training a learning model with pairs of intensity distributions and light correction coefficients before actual aberration correction. This pre-processing creates a ready-to-use predictive system that can quickly provide correction coefficients without performing full calculations during actual correction operations, thus reducing calculation time while maintaining accuracy
Solution Approach 2:
The patent uses copying by creating a learned model that replicates the relationship between intensity distributions and correction coefficients. Instead of recalculating from scratch each time, the system copies the essential correction patterns learned during training and applies them to new cases, significantly reducing computational burden while preserving correction accuracy
2Loss of information
If full intensity image data is used as input for prediction, then comprehensive information is available, but the data amount and processing load increase
Solution Approach 1:
The patent applies the extraction principle by selecting and extracting only the most relevant features from the intensity image data. Specifically, it extracts intensity distributions from multiple regions of interest and flattens them into feature vectors, removing redundant information while preserving the essential characteristics needed for accurate correction coefficient prediction
Solution Approach 2:
The patent applies segmentation by dividing the intensity image into multiple regions of interest and processing each region separately. This segmentation allows the system to focus on specific areas that contain the most relevant information for aberration correction, reducing the overall data volume while maintaining predictive accuracy
Data Source
AI summary
A control device includes: an acquisition unit that acquires, for an intensity image obtained by observing an action caused by light corrected using a spatial light modulator based on a Zernike coefficient, an intensity distribution that is a distribution of intensities in a plurality of regions of interest within a predetermined range on the intensity image; a generation unit that calculates a comparison result between the intensity distribution and a target distribution to generate comparison data; and a prediction unit that predicts a Zernike coefficient, which is for performing aberration correction related to the light so that the intensity distribution approaches the target distribution, by inputting the comparison data and the Zernike coefficient, which is a basis of the intensity distribution, to a learning model.