Novel Outlier Filtering Method for AOI Image Databases

Takács, Tibor; Vajta, László

doi:10.1109/tcpmt.2012.2184765

Cited by 10 publications

(6 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…1) It is obviously important to improve the accuracy of PCB abnormal detection in AOI system. To this end, Takacs et al present a method to find and remove the outlier images from the database [27]. However, the proposed system cannot realize a stable performance under different scenarios.…”

Section: Machine Learning and Industrial Iotmentioning

confidence: 99%

Improving accuracy of automatic optical inspection with machine learning

Tong

Tian

et al. 2021

Front. Comput. Sci.

View full text Add to dashboard Cite

Electronic devices require the printed circuit board (PCB) to support the whole structure, but the assembly of PCBs suffers from welding problem of the electronic components such as surface mounted devices (SMDs) resistors. The automated optical inspection (AOI) machine, widely used in industrial production, can take the image of PCBs and examine the welding issue. However, the AOI machine could commit false negative errors and dedicated technicians have to be employed to pick out those misjudged PCBs. This paper proposes a machine learning based method to improve the accuracy of AOI.In particular, we propose an adjacent pixel RGB value based method to pre-process the image from the AOI machine and build a customized deep learning model to classify the image. We present a practical scheme including two machine learning procedures to mitigate AOI errors. We conduct experiments with the real dataset from a production line for three months, the experimental results show that our method can reduce the rate of misjudgment from 0.3%-0.5% to 0.02%-0.03%, which is meaningful for thousands of PCBs each containing thousands of electronic components in practice.

show abstract

Section: Machine Learning and Industrial Iotmentioning

confidence: 99%

Improving accuracy of automatic optical inspection with machine learning

Tong

Tian

et al. 2021

Front. Comput. Sci.

View full text Add to dashboard Cite

show abstract

“…11 Moreover, AOI is limited by the need to make constant adjustments according to changing quality status. These adjustments are necessary to maintain a low false call rate when performing quality checks; if not appropriately adjusted, the inspection effectiveness may be deteriorated and the process delayed [12][13] . The existing AOI process depends on quality engineers to calibrate inspection procedures and classify defective goods, which may lead to severe errors.…”

Section: Introductionmentioning

confidence: 99%

A System Deployment Model of Multi-CCD Automatic Optical Inspection for Economical Operations

Wang

Qiu

2022

IEEE Access

View full text Add to dashboard Cite

Overkill and leakage are common problems in automated optical inspection (AOI) systems that have harmed firms to devote in intensive manual reinspection and seek the best CCD setting, resulting in more costly and less effective AOI operations. This study proposed a two-stage AOI based inspection system deployment model of multi-charge coupled device (CCD) for cost reduction; that is, the strictness decision model of a single CCD in the operational stage and switch on/off decision model of a set of CCDs in the strategic stage. The strictness decision model identifies the best strictness level of individual CCD by analyzing a confusion matrix according to the true defect rate of products. The multi-CCD switch decision model suggests the best switch on/off decision of multiple CCDs. Experiments in an electrical connector manufacturing plant revealed that the total inspection cost was the lowest when the strictness level was set close to the true defect rate. The best multi-CCD on/off strategy was more cost-effective than was the all CCDs-on mode under various strictness settings, by cost reduction up to 23.49%. In general, the proposed two-stage model achieves flexible multi-CCD settings of AOI systems and operations at a low cost. The proposed model can be extended to different applications of overkill and leakage problems. In this study, the proposed VGG-like convent five-layer convolution model learns from the experience of quality re-inspection personnel, which can be applied to the re-inspection line to identify AOI as defective images for re-inspection. The best accuracy reaches 99.21%.INDEX TERMS automated optical inspection, charge coupled device, confusion matrix, cost-effective analysis, inspection system deployment, Type I & II errors, CNN.

show abstract

“…To research solder defect detection systems based on AOI of electronic components, the key issue is to solve the problems of image feature extraction of AOI, defect classification and algorithm running efficiency (Saenthon and Kaitwanidvilai, 2010; Jiang et al , 2012; Takacs and Vajta, 2012; Wu et al , 2013). To solve these problems, a series of studies have been conducted, mainly in the following areas: Image feature extraction and selection : The main features are the gray values (Kim and Cho, 1995; Kim et al , 1999; Ong et al , 2008; Lin et al , 2007), the histogram (Sankaran et al , 1995), the wavelet feature (Belbachir et al , 2005), the geometric characteristics (Acciani et al , 2006), texture (Mar et al , 2012) and so on.…”

Section: Introductionmentioning

confidence: 99%

“…Wherein, Ong et al (2008) made use of a supervised self-learning neural network (LVQ), the input parameters were the image of the gray values and the introduction of a competition mechanism within the network. The output was of three kinds: insufficient solder, excessive solder and normal. Using mixed vertical light and side light to obtain images, the results showed that the detection accuracy had been improved compared to single vertical light results, but the light source was too complicated and not suitable for large-scale real-time needs. Classification of solder joint defects using an unsupervised learning model : Such classification algorithms are mainly self-organizing neural networks (Ko and Cho, 2000) and unsupervised clustering (Takacs and Vajta, 2012). Ko and Cho (2000) made use of an unsupervised self-organization neural network and fuzzy set theory.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Solder joint defect classification based on ensemble learning

Wu¹

2017

SSMT

View full text Add to dashboard Cite

Purpose This paper aims to inspect the defects of solder joints of printed circuit board in real-time production line, simple computing and high accuracy are primary consideration factors for feature extraction and classification algorithm. Design/methodology/approach In this study, the author presents an ensemble method for the classification of solder joint defects. The new method is based on extracting the color and geometry features after solder image acquisition and using decision trees to guarantee the algorithm’s running executive efficiency. To improve algorithm accuracy, the author proposes an ensemble method of random forest which combined several trees for the classification of solder joints. Findings The proposed method has been tested using 280 samples of solder joints, including good and various defect types, for experiments. The results show that the proposed method has a high accuracy. Originality/value The author extracted the color and geometry features and used decision trees to guarantee the algorithm's running executive efficiency. To improve the algorithm accuracy, the author proposes using an ensemble method of random forest which combined several trees for the classification of solder joints. The results show that the proposed method has a high accuracy.

show abstract

Novel Outlier Filtering Method for AOI Image Databases

Cited by 10 publications

References 11 publications

Improving accuracy of automatic optical inspection with machine learning

Improving accuracy of automatic optical inspection with machine learning

A System Deployment Model of Multi-CCD Automatic Optical Inspection for Economical Operations

Solder joint defect classification based on ensemble learning

Contact Info

Product

Resources

About