Periodic Pattern Defect Detection Using Slope Angle Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing defect detection methods for images with periodic patterns are inefficient due to high calculation time and inaccuracy when dealing with complex patterns, as they require repeated shifting to measure minimum error, leading to prolonged processing times and inaccurate pattern period extraction.
Innovation Solution
A defect detection apparatus and method that extracts the slope angle of a periodic pattern, uses this angle to determine the pattern period, and shifts the image accordingly to enhance arithmetic operation speed and accurately detect defects by removing overlapping areas between original and shifted images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If repeated shifting is performed to extract pattern period by measuring minimum error, then pattern period can be extracted, but calculation time is excessively long
Solution Approach 1:
The patent applies preliminary action by first extracting the slope angle of the periodic pattern before performing image shifting. By pre-determining the slope angle, the system can directly calculate the optimal shifting amount without requiring repeated trial shifts, thus reducing calculation time while maintaining accurate pattern period extraction
Solution Approach 2:
The patent changes the parameter approach from brute-force repeated shifting to a mathematical calculation method. By extracting the slope angle and using it to compute the shifting amount directly, the system transforms a time-consuming iterative process into a efficient single-step calculation, resolving the contradiction between accuracy and time
2Measurement precision
If repeated shifting is performed to extract pattern period, then pattern period can be extracted, but processing complexity increases
Solution Approach 1:
The patent simplifies the arithmetic operation complexity by changing from multiple repeated shifting operations to a single slope angle extraction followed by a direct calculation. This parameter transformation reduces the number of arithmetic operations while maintaining extraction accuracy
3Ease of operation
If related art defect detection method is applied to image with complicated shaped pattern, then defect detection can be performed, but pattern period extraction becomes inaccurate
Solution Approach 1:
The patent addresses complicated shaped patterns by changing the approach from error-based minimum measurement to slope angle-based direct calculation. The slope angle extraction method is more robust to pattern shape variations, maintaining accuracy even when patterns have complex geometries that make error measurement unreliable
Data Source
AI summary
Disclosed is a defect detection apparatus. The defect detection apparatus includes an angle extractor configured to extract a slope angle of a pattern from an original image in which a plurality of the patterns are formed at periodic intervals, a pattern period extractor configured to extract a pattern period at which the patterns are separated from each other, by using the slope angle, and an image shifter configured to shift the original image by the pattern period in a direction perpendicular to the slope angle to form the shifted image. The present invention shifts an image by using the slope angle and pattern period of the periodic pattern, thus easily extracting a defect of the original image.