Semiconductor Package Marking Inspection Using Reference Image Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional marking inspection methods for semiconductor packages are inefficient and inaccurate, often misclassifying normal packages as defective or vice versa, due to the need to recognize and compare character strings for all packages, leading to prolonged inspection times and reduced efficiency.
Innovation Solution
A method and apparatus that acquire character string information from a target semiconductor package, store it as reference information, and compare image information from subsequent packages to quickly determine defective markings, using a camera, input unit, memory, display, controller, alarm, and classification module to classify packages accurately.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If optical character recognition is used to recognize character strings of all semiconductor packages, then marking inspection accuracy is improved, but inspection time is significantly prolonged
Solution Approach 1:
The patent applies preliminary action by capturing and storing reference image information from the first package of each lot before inspection begins. This pre-established reference is then used for rapid comparison with subsequent packages, eliminating the need to perform full optical character recognition on every package while maintaining inspection accuracy.
Solution Approach 2:
The patent uses copying by creating a reference image copy from the first package and storing it in memory. This reference copy is then repeatedly compared against images of subsequent packages, replacing the need for repeated full character recognition while preserving the ability to detect marking defects.
2Productivity
If the first package is used as reference for all subsequent packages in a lot, then inspection speed is improved, but inspection accuracy deteriorates when the first package has defective markings
Solution Approach 1:
The patent implements feedback by continuously monitoring package inspection results and automatically updating the reference image when a normal package is detected. When defective packages are encountered, the system maintains the existing reference, ensuring that defective markings do not propagate as the new reference standard.
Solution Approach 2:
The patent applies dynamics by making the reference image adaptable and changeable during the inspection process. Instead of fixing the reference to the first package, the system dynamically updates the reference when normal packages are identified, allowing the inspection standard to evolve based on actual product quality.
3Measurement precision
If character string recognition is performed on every package, then marking defects are accurately detected, but inspection efficiency is reduced
Solution Approach 1:
The patent extracts only the essential reference information from the first package and stores it for comparison, removing the need to perform full character recognition on every subsequent package. This extraction approach maintains defect detection capability while dramatically improving inspection efficiency.
Solution Approach 2:
The patent applies partial action by performing full character recognition only once on the first package to establish the reference, then using simpler image comparison for subsequent packages. This partial approach to recognition maintains sufficient accuracy for defect detection while significantly reducing overall inspection time.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach significantly enhances the speed and accuracy of marking inspections by only comparing image information after establishing reference character string information, reducing errors and improving overall inspection efficiency.
Implementation Method 1
captures a marking character of a first package of a new lot using a CCD camera to detect an image of the marking character
Data Source
Figure 1
Figure 2
AI summary
A method and apparatus for inspecting a marking of a semiconductor package is disclosed, which quickly carries out a marking inspection of a semiconductor package, and increases the accuracy of the inspection. The method includes: a) acquiring character string information of a semiconductor package of a target lot to be inspected, and storing the acquired character string information as reference character string information; b) capturing a character pattern printed on an incoming semiconductor package, and acquiring image information of the incoming semiconductor package/ c) displaying the acquired image information to allow a user to visually recognize the acquired image information; d) detecting character string information from the displayed image information according to a setup signal, determining the displayed semiconductor-package image information to be reference image information of a corresponding lot if the detected character string information is equal to the pre- stored character string information; and e) classifying a corresponding semiconductor package into a defective-package storage unit if the determined reference image information is different from image information of sequentially-incoming semiconductor packages.