Automated optical inspection (AOI) image classification method, system and computer-readable media

• US 10,636,133 B2
• Filed: 12/19/2017
• Issued: 04/28/2020
• Est. Priority Date: 10/27/2017
• Status: Active Grant

• Current Assignee: Industrial Technology Research Institute
• Original Assignee: Industrial Technology Research Institute
• Inventors: Chang, Keng-Hao, Chiu, Wei-Yao, Tsai, Ya-Hui, Hu, Jwu-Sheng
• Primary Examiner(s): Hsieh, Ping Y
• Application Number:
  US15/846,886
• Publication Number:
  US 20190130555A1
• Time in Patent Office: 861 Days
• CPC Class Codes: G06F 18/22 Matching criteria, e.g. proximity measures
• CPC Class Codes: G06F 18/2413 based on distances to training or reference patterns
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30108 Industrial image inspection
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: G06V 10/454 Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]
• CPC Class Codes: G06V 10/761 Proximity, similarity or dissimilarity measures
• CPC Class Codes: G06V 2201/06 Recognition of objects for industrial automation
• CPC Class Codes: G06V 30/19173 Classification techniques

Abstract

An automated optical inspection (AOI) image classification method includes sending a plurality of NG information of a plurality of samples from an AOI device into an Artificial Intelligence (AI) module; performing discrete output calculation on the NG information of the samples by the AI module to obtain a plurality of classification information of the samples; performing kernel function calculation on the classification information of the samples by the AI module to calculate respective similarity distances of the samples and performing weighting analysis; based on weighting analysis results of the samples, judging classification results of the samples; and based on the classification results of the samples, performing classification of the samples.

Claims

1. An automated optical inspection (AOI) image classification method including:

feeding a plurality of NG information of a plurality of samples from an AOI device into an artificial intelligence (AI) training module;

performing discrete output on the plurality of NG information of the samples by the AI training module to generate a plurality of classification information of the samples;

performing kernel function on the plurality of classification information of the samples by the AI training module to calculate respective similarity distances of the samples and to perform weighting analysis;

performing classification determination based on weight analysis results of the samples, to determine respective classification results of the samples; and

based on the respective classification results of the samples, classifying the samples;

wherein the step of calculating respective similarity distances of the samples includes;

calculating a plurality of similarity distances between the classification information of the samples and a plurality of ideal classification.

2. The method according to claim 1, wherein the step of perform weighting analysis includes:

assigning respective weighting to the similarity distances of the samples, wherein if a first similarity distance of a first sample is smaller than a second similarity distance of the first sample, a first weight assigned to the first similarity distance of the first sample is higher than a second weight assigned to the second similarity distance of the first sample.

3. The method according to claim 1, wherein the classification results of the samples are fed back to the AI training module for training the AI training module.

4. The method according to claim 1, wherein the step of classifying the samples includes:

classifying by a classification device including any combination of a pneumatic cylinder, a conveyor, a cantilever mechanism, a robot arm, and a carrier.

5. A non-transitory computer-readable media, when loaded by an automated optical inspection (AOI) image classification system, the AOI image classification system executing the AOI image classification method in claim 1.

7. The system according to claim 1, wherein the AI training module assigns respective weighting to the similarity distances of the samples, wherein if a first similarity distance of a first sample is smaller than a second similarity distance of the first sample, a first weight assigned to the first similarity distance of the first sample is higher than a second weight assigned to the second similarity distance of the first sample.

6. An automated optical inspection (AOI) image classification system including:

an AOI device, for performing automated optical inspection on a plurality of samples to obtain respective OK information or NG information of the samples; and

an artificial intelligence (AI) training module coupled to the AOI device, the AI training module receiving the plurality of NG information of the samples from the AOI device, the AI training module performing discrete output on the NG information of the samples to generate a plurality of classification information of the samples, the AI training module performing kernel function on the plurality of classification information of the samples to calculate respective similarity distances of the samples and to perform weighting analysis, and the AI training module performing classification determination based on weight analysis results of the samples, to determine respective classification results of the samples;

wherein the AI training module calculates a plurality of similarity distances between the classification information of the samples and a plurality of ideal classification.

8. The system according to claim 6, wherein the classification results of the samples are fed back to the AI training module for training the AI training module.

9. The system according to claim 6, wherein the classification results, obtained by the AI training module, of the samples are classified by a classification device including any combination of a pneumatic cylinder, a conveyor, a cantilever mechanism, a robot arm, and a carrier.