Inspection Apparatus Alignment Using Least-Squares Exclusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional pattern inspection apparatuses face challenges in achieving highly precise alignment between reference and optical images, especially with miniaturization of patterns, leading to difficulties in detecting ultrafine defects due to systematic errors and local defects affecting alignment calculations.
Innovation Solution
A target workpiece inspection apparatus and method that includes an optical image acquiring unit, a reference image generating unit, a difference judging unit, a least-squares method displacement calculating unit, and a position correcting unit to perform precise positional alignment by calculating displacement amounts using a regular matrix, excluding pixels with large defects to prevent excessive alignment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional least-squares method is used for alignment calculation, then alignment can be performed, but systematic errors and local defects cause excessive alignment and reduce measurement precision
Solution Approach 1:
The patent extracts and removes pixels with large defects from the alignment calculation process. By identifying pixels where the absolute difference between reference image and optical image exceeds a threshold, these problematic pixels are excluded from the least-squares calculation, preventing them from causing excessive alignment and improving both measurement precision and reliability
Solution Approach 2:
The patent applies different processing quality to different regions of the image. Pixels are classified into two categories: those with small differences are used for alignment calculation, while those with large differences are excluded. This local differentiation ensures that only reliable pixel data contributes to the alignment calculation, resolving the contradiction between precision and reliability
2Quantity of substance
If all pixels are used in alignment calculation, then more data is available for calculation, but pixels with large defects cause excessive alignment
Solution Approach 1:
The patent extracts pixels with large defects based on a threshold criterion and removes them from the calculation set. This extraction process ensures that only pixels with acceptable quality (small absolute difference between images) are used, maintaining data quantity while improving alignment accuracy by excluding problematic pixels
Solution Approach 2:
The patent changes the parameter selection criterion for pixels used in calculation. Instead of using all pixels uniformly, it introduces a threshold parameter that dynamically determines which pixels qualify for inclusion. This parameter-based filtering optimizes the balance between data quantity and calculation accuracy
Data Source
AI summary
A target workpiece inspection apparatus comprises an optical image acquiring unit to acquire an optical image of a target workpiece, a reference image generating unit to generate a reference image to be compared, a difference judging unit to judge whether an absolute value of difference between pixel values of the images in each pixel at a preliminary alignment position between the images is smaller than a threshold value, a least-squares method displacement calculating unit to calculate a displacement amount displaced from the preliminary alignment position, by using a regular matrix for a least-squares method obtained from a result judged, a position correcting unit to correct an alignment position between the optical image and the reference image to a position displaced from the preliminary alignment position by the displacement amount, and a comparing unit to compare the optical image and the reference image whose alignment position has been corrected.