GAN-Based Semiconductor Package Mark Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing miniaturization of semiconductor packages makes visual inspection time-consuming and costly, leading to potential misjudgments and distribution of defective products.
Innovation Solution
A method using a deep learning model with a generative adversarial network (GAN) to generate fake images for inspecting marks on semiconductor packages, comparing these fake images with measured images to determine defects such as misprinting, cutting, blur, noise, displacement, double printing, reverse printing, bay, and thermochromic phenomena.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If visual inspection is used for semiconductor package marks, then inspection can be performed, but inspection time increases and cost increases due to miniaturization
Solution Approach 1:
The system performs preliminary action by pre-processing the measured image to extract mark region information and generating a region of interest (ROI) image before comparison. This preliminary extraction and focusing on relevant regions accelerates the inspection process without sacrificing accuracy, directly addressing the time loss issue caused by miniaturization
Solution Approach 2:
The system creates a reference mark image (copy of the expected correct mark) and compares it with the measured mark image. This copying approach enables automated optical inspection that is both fast and accurate, resolving the contradiction between inspection time and measurement precision by using image replication rather than manual inspection
2Measurement precision
If visual inspection is used for semiconductor package marks, then inspection can be performed, but cost increases due to miniaturization
Solution Approach 1:
The system replaces manual visual inspection (mechanical human operation) with an automated optical inspection system using image processing and comparison algorithms. This substitution eliminates labor costs and reduces operational expenses while maintaining high inspection accuracy, directly addressing the cost increase issue
Solution Approach 2:
By using reference mark images and automated image comparison, the system eliminates the need for expensive manual inspection processes. The copying and digital comparison approach is significantly more cost-effective while maintaining measurement precision
3Ease of operation
If manual inspection is used, then flexibility is maintained, but misjudgment occurs leading to distribution of defective products
Solution Approach 1:
The system implements feedback by comparing the measured mark image with the reference mark image and automatically determining whether the mark is normal or defective. This automated feedback mechanism eliminates human misjudgment while maintaining operational flexibility through programmable inspection criteria
Solution Approach 2:
The inspection system performs self-service by automatically processing images, comparing marks, and determining defect status without human intervention. This self-automated process both maintains flexibility through configurable parameters and eliminates misjudgment, resolving the reliability issue
Data Source
AI summary
Provided is a mark inspection method of a semiconductor package. The method includes receiving a measured image of a semiconductor package comprising a mark including at least one of a figure and a character string, receiving a standard defect image corresponding to the measured image, the standard defect image includes the mark having a defect comprising at least one of misprinting, cutting, blur, noise, displacement, double printing, reverse printing, bay, and thermochromic phenomena, generating a fake image through a deep learning model using a generative adversarial network (GAN) with the measured image and the standard defect image, and comparing the fake image with the measured image.