Sample Localization with Coarse Image Regions for Cover Glass Edges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for identifying cover glass edges in high-resolution images are resource-intensive, time-consuming, and prone to errors, particularly in low-contrast recordings, leading to poor customer acceptance.
Innovation Solution
A method utilizing a selection model to identify structure regions in coarse image data, followed by an identification model to determine localization, reducing data detail and computational resources while maintaining accuracy through machine learning models like classifiers and segmentation models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional image analysis tools are used to identify cover glass edges in high-resolution recordings, then complete image analysis is performed, but the analysis becomes resource-intensive, time-consuming and data-intensive
Solution Approach 1:
The patent divides the high-resolution image data into multiple resolution levels (coarse and fine). The selection model first identifies structure regions in coarse image data, then the identification model processes only these relevant regions in high resolution. This segmentation of processing by resolution level reduces computational resources and time while maintaining identification accuracy.
Solution Approach 2:
The patent extracts only the relevant structure regions from the complete high-resolution image data for detailed analysis. By using the selection model to identify and extract these specific regions based on coarse image data, the system avoids processing the entire high-resolution dataset, thereby reducing data intensity and processing time while focusing computational resources on areas most likely to contain cover glass edges.
2Measurement precision
If a single high-resolution recording is analyzed with conventional tools, then detailed structure information is obtained, but the analysis is flawed and has poor customer acceptance
Solution Approach 1:
The patent performs preliminary analysis using the selection model on coarse image data before conducting detailed analysis on high-resolution data. This preliminary step identifies structure regions that are likely to contain relevant features, allowing the subsequent detailed analysis to focus on these pre-identified areas. This two-stage approach improves reliability by ensuring that detailed analysis is applied where it is most needed, rather than uniformly across all data.
Solution Approach 2:
The patent introduces coarse image data as an intermediary between the raw high-resolution data and the final identification results. The coarse image data serves as a mediator that guides the selection of relevant regions, enabling the system to bridge the gap between comprehensive high-resolution data and focused analysis, thereby improving the reliability of detection results.
3Measurement precision
If elastic grid deformation is applied to maintain resolution in relevant regions, then intermediate image is created, but geometric distortions occur and deformations cannot be interpreted
Solution Approach 1:
Instead of applying elastic grid deformation to the entire image, the patent segments the image into coarse and fine resolution levels. The selection model operates on coarse data to identify structure regions, and only these identified regions are processed in high resolution. This segmentation approach maintains geometric integrity while preserving necessary resolution in relevant areas.
Solution Approach 2:
The patent creates a coarse copy of the high-resolution image data for the purpose of identifying structure regions. This coarse copy serves as a simplified representation that guides the selection process without requiring geometric transformations of the original high-resolution data, thereby avoiding distortion issues while still enabling region identification.
Data Source
AI summary
The present disclosure relates to a method for determining a localization of a sample based on structure information, wherein elements of a sample holding device holding the sample in an imaging device form structures in image data captured with the imaging device, comprising providing the image data, determining, by means of a selection model, structure regions based on coarse image data, determining, by means of an identification model, the structure information based on the structure regions, and determining a localization of the sample based on the structure information, characterized in that the coarse image data are reduced in detail compared to the image data, the structure regions are regions in the image data in which the structures are captured with a certain probability and a sum of data quantities of the structure regions is smaller than the data quantity of the image data.