Product Defect Detection Using Autoencoder Difference Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing defect detection methods struggle to accurately distinguish between background errors and real defects, leading to low accuracy in defect detection.
Innovation Solution
A method utilizing a pre-trained autoencoder to generate a reconstructed image, a difference image, and perform clustering processing to identify feature absolute values, followed by generating a target image and determining defect detection results based on these values, effectively distinguishing between background errors and defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional defect detection methods are used, then the detection process is simple, but the accuracy of distinguishing between background errors and real defects is low
Solution Approach 1:
The patent segments the image processing into distinct stages: obtaining the image, generating difference image through autoencoder, performing clustering processing on difference image, and generating target image. This segmentation allows each stage to focus on specific tasks, improving overall detection accuracy while maintaining manageable process complexity
Solution Approach 2:
The patent introduces an intermediary clustering processing step that analyzes the difference image to identify feature absolute values. This intermediary step acts as a mediator between the raw difference image and the final defect detection, filtering out background errors while preserving real defects through mathematical clustering operations
2Measurement precision
If background errors are not filtered out, then the detection process remains simple, but the distinction between defect-indicating pixel values and background noise is poor
Solution Approach 1:
The patent extracts and removes background errors from the difference image through clustering processing. By identifying and separating the clustering centers corresponding to background errors from those corresponding to real defects, the method extracts only the relevant defect information while discarding background noise, achieving clear distinction between defects and background
Solution Approach 2:
The patent changes the parameter representation by transforming the image into a difference image with adjusted pixel values, then further processing through clustering to create target image with modified feature absolute values. These parameter changes enable better differentiation between defect and background regions that were not distinguishable in the original image
Data Source
AI summary
A method for detecting a product for defects implemented in an electronic device, the method obtains an image of a product to be detected, obtains a reconstructed image by inputting the image to be detected into a pre-trained autoencoder, generates a difference image from the image and the reconstructed image, obtains a number of feature absolute values by performing clustering processing on the difference image; generates a target image according to the number of feature absolute values, the difference image, and a preset value; and determines a defect detection result by detecting the target image for defects.