Open AccessCNN‐based reference comparison method for classifying bare PCB defects
Author(s) -
Wei Peng,
Liu Chang,
Liu Mengyuan,
Gao Yunlong,
Liu Hong
Publication year - 2018
Publication title -
the journal of engineering
Language(s) - English
Resource type - Journals
ISSN - 2051-3305
DOI - 10.1049/joe.2018.8271
Subject(s) - printed circuit board , convolutional neural network , computer science , artificial intelligence , process (computing) , set (abstract data type) , pattern recognition (psychology) , image (mathematics) , artificial neural network , image processing , computer vision , programming language , operating system
Printed circuit board (PCB) inspection is an essential part of PCB production process. Traditional PCB bare board defect detection methods have their own defects. However, the PCB bare board defect detection method based on automatic optic inspection is a feasible and effective method, and it is having more and more application in industry. Based on the idea of the reference comparison method, this study aims at studying the classification of defects. First of all, the method of extracting defect areas using morphology is studied; meanwhile, a data set containing 1818 images with 6 different detailed defect area image parts are produced. Then, in order to classify defects accurately, a traditional classification algorithm based on digital image processing was attempted, and a defect classification algorithm based on convolutional neural network was proposed. After experimental demonstration, in the actual results, the defect classification algorithm based on convolutional neural network can achieve a fairly high classification accuracy (95.7%), which is much higher than the traditional method, and the new method has stronger stability than the traditional one.