Multi-Channel Image Defect Detection for Conflicting Inspection Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current defect detection methods in manufactured articles, such as semiconductor substrates and batteries, are inefficient in reconciling conflicting data from images taken under different attributes, leading to inaccurate identification and classification of defects, which can result in scrapped products and significant costs.
Innovation Solution
A multi-channel image processing system using Artificial Intelligence and machine learning models, specifically a multi-channel neural network, is employed to construct and analyze images with varying attributes, such as lighting conditions and angles, to determine the presence and classification of defects in manufactured articles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple images with different imaging attributes are used for defect detection, then the detection accuracy can be improved, but the data conflict and reconciliation complexity increases
Solution Approach 1:
The patent combines multiple images with different imaging attributes (lighting conditions, angles, magnifications) into a unified multi-channel image representation. Each image becomes a channel in the multi-channel image, allowing the neural network to process and reconcile all image data simultaneously through learned feature fusion, rather than requiring separate reconciliation steps for conflicting data.
Solution Approach 2:
The multi-channel image serves as an intermediary representation that bridges the gap between multiple source images and the defect detection task. This intermediate structure organizes the complex multi-attribute image data in a standardized format that the neural network can efficiently process, acting as a mediator between the diverse image inputs and the detection algorithm.
2Measurement precision
If traditional image processing methods are used, then the processing speed is fast, but the defect identification accuracy is insufficient
Solution Approach 1:
The patent replaces traditional mechanical image processing methods (filtering, edge detection, thresholding) with an artificial neural network-based system. The neural network learns optimal defect detection features directly from the multi-channel image data, substituting manual feature engineering and sequential processing steps with a unified learned model that achieves both high accuracy and efficient processing.
Solution Approach 2:
The system changes the parameter representation by transforming multiple images with different attributes into a multi-channel image structure. This parameter transformation allows the neural network to simultaneously consider variations in lighting, angle, and magnification as distinct channels, enabling more accurate defect identification without requiring multiple separate processing pipelines.
3Reliability
If single-channel image analysis is used, then the processing is simple, but the defect detection reliability is low
Solution Approach 1:
The patent transitions from single-channel (2D) image analysis to multi-channel (3D+) image analysis by adding the channel dimension. Each image with different imaging attributes becomes a separate channel, creating a multi-dimensional data structure that enables the neural network to detect defects from multiple perspectives simultaneously, significantly improving detection reliability through dimensional expansion.
Data Source
AI summary
A system and/or method of detecting a defect in a manufactured article. Images of the manufactured article are generated having different imaging attributes from one another. A multi-channel image is constructed using the images, each image channel of the multi-channel image corresponding to one of the images. The multi-channel image is input to a processor, which determines, based on the input. at least whether the manufactured article includes a defect.