Image Restoration with Deformation Correction for Wafer Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image restoration techniques for semiconductor wafers face challenges in creating high-quality images due to imaging damage from electron beam irradiation, deformation, visual field deviation, and luminance changes, making it difficult to generate appropriate supervision for fine circuit patterns.
Innovation Solution
An image restoration system that includes an image restoration unit, a deformation prediction unit, and a deformation correction unit, which predict and correct deformation between low-quality images to generate accurate prediction images, and an image restoration method that trains using these corrected images to reduce loss function evaluation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple imaging is performed to accumulate imaging results and create high quality images, then image quality and accuracy are improved, but throughput decreases due to increased imaging time
Solution Approach 1:
The system performs preliminary deformation prediction and correction on individual low-quality images before they are accumulated. By predicting deformation amounts using the deformation prediction unit and correcting images using the deformation correction unit, the system prepares images in advance so that they can be properly accumulated without requiring excessive imaging times, thus maintaining throughput while improving image quality
Solution Approach 2:
The deformation prediction unit and deformation correction unit act as intermediary components between the raw low-quality images and the final accumulated high-quality images. These units predict and correct deformation amounts, serving as a bridge that enables effective image accumulation without requiring excessive imaging operations, thereby resolving the contradiction between image quality and throughput
2Measurement precision
If multiple imaging is performed to create high quality images for supervision, then image restoration accuracy is improved, but imaging damage and deformation occur due to electron beam irradiation
Solution Approach 1:
The system converts the harmful effect of electron beam irradiation into a beneficial process by using the deformation prediction unit to predict deformation amounts caused by imaging damage. Instead of avoiding multiple imaging operations, the system embraces the deformation as predictable and correctable, using the deformation correction unit to compensate for imaging damage, thus transforming harm into benefit
Solution Approach 2:
The system implements a feedback mechanism where the deformation prediction unit continuously predicts deformation amounts based on imaging results, and the deformation correction unit applies corrections based on these predictions. This closed-loop feedback allows the system to account for and compensate imaging damage without requiring excessive additional imaging, thereby improving image restoration accuracy while limiting further imaging damage
3Ease of manufacture
If simple average image is used for supervision, then processing is simplified, but appropriate supervision cannot be created due to visual field deviation and luminance change
Solution Approach 1:
The deformation correction unit serves as an intermediary between simple averaging and complex processing. It applies deformation correction to individual images before they are averaged, ensuring that visual field deviation and luminance changes are compensated. This approach maintains processing simplicity while significantly improving supervision accuracy compared to direct averaging of uncorrected images
Data Source
AI summary
An image quality improvement system includes: an image quality improvement unit that improves the image quality of a low quality image; a deformation prediction unit that predicts a deformation amount that has occurred between a first low quality image and a different second low quality image, included in a series of input low quality images; and a deformation correction unit that corrects, based on the deformation amount predicted by the deformation prediction unit, one of a first prediction image obtained by applying processing by the image quality improvement unit to the first low quality image, the second low quality image, or a second prediction image obtained by applying processing by the image quality improvement unit to the second low quality image. The image quality improvement system learns to reduce the evaluation of a loss function between the first prediction and the second low quality image or the second prediction image.