Surface Defect Size Detection for Real-Time Modification Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated quality assurance methods for surface modification processes, such as laser beam brazing and welding, face challenges in accurately determining the size of defects in real-time with high throughput, leading to inefficiencies and increased computational requirements.
Innovation Solution
A computer-implemented method using a YOLO-style model for defect size determination, where defect occurrence is identified separately from size assessment, allowing for high-speed defect identification and accurate size classification with reduced computational effort, utilizing a trained neural network and image classification to verify defect presence across multiple image frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated quality assurance methods using image acquisition and analysis are used, then productivity is improved, but measurement precision deteriorates due to difficulty in accurately determining defect sizes
Solution Approach 1:
The patent segments the defect analysis process into two distinct stages: first identifying whether a defect is present, then separately determining its size characteristics. This segmentation allows each stage to be optimized independently, with the size determination stage focusing computational resources only on regions where defects have been identified, thereby maintaining high measurement precision while preserving automated throughput.
Solution Approach 2:
The patent performs preliminary defect identification before conducting detailed size measurements. By first detecting the presence of defects through image acquisition and preliminary analysis, the system prepares a targeted list of regions requiring detailed measurement, thus avoiding unnecessary computational effort on defect-free areas and maintaining both speed and precision.
2Measurement precision
If detailed defect size determination is performed for all detected defects, then measurement precision is improved, but use of energy increases due to high computational requirements
Solution Approach 1:
The patent applies partial action by performing comprehensive size determination only for regions where defects have been identified, rather than analyzing every region in the image. This selective approach ensures high measurement precision for actual defects while avoiding the excessive computational energy consumption that would result from processing the entire image at full resolution and detail.
Solution Approach 2:
The patent applies local quality by allocating different levels of analysis depth to different regions of the image. Regions identified as containing defects receive detailed size determination with high computational resources, while regions without defects receive minimal or no detailed analysis. This localized quality approach maintains measurement precision where needed while minimizing overall energy consumption.
3Productivity
If high-speed image acquisition is used to maintain productivity, then productivity is improved, but measurement precision deteriorates due to reduced image quality and increased false positives
Solution Approach 1:
The patent segments the image processing into multiple passes: a first pass using lower-resolution or lower-frame-rate imaging for rapid defect detection, followed by a second pass that applies detailed size determination only to identified defect regions. This segmentation allows the system to maintain high productivity through fast initial detection while achieving high measurement precision in the subsequent detailed analysis stage.
Solution Approach 2:
The patent performs preliminary defect detection using high-speed image acquisition to identify regions of interest, then follows up with more detailed analysis of those specific regions. This preliminary action approach allows the system to maintain high productivity through rapid initial screening while ensuring measurement precision through subsequent detailed examination of only the necessary regions.
Data Source
AI summary
A method is specified for determining a size of a defect occurring in a surface region of a component while a surface modification process is performed on the surface region. The method includes identifying an occurrence of a defect occurring in a surface region of a component on a basis of a set of images and determining a size of the defect in a separate method step from the occurrence of the defect identified. In addition, an apparatus and a computer program are specified for determining a size of a defect occurring in a surface region of a component while a surface modification process is performed on the surface region.