A Review on Recent Advances in Vision-based Defect Recognition towards Industrial Intelligence

Yiping Gao,Xinyu Li,Xi Vincent Wang,Lihui Wang,Liang Gao

DOI: https://doi.org/10.1016/j.jmsy.2021.05.008

IF: 12.1

2022-01-01

Journal of Manufacturing Systems

Abstract:<p>In modern manufacturing, vision-based defect recognition is an essential technology to guarantee product quality, and it plays an important role in industrial intelligence. With the developments of industrial big data, defect images can be captured by ubiquitous sensors. And, how to realize accuracy recognition has become a research hotspot. In the past several years, many vision-based defect recognition methods have been proposed, and some newly-emerged techniques, such as deep learning, have become increasingly popular and have addressed many challenging problems effectively. Hence, a comprehensive review is urgently needed, and it can promote the development and bring some insights in this area. This paper surveys the recent advances in vision-based defect recognition and presents a systematical review from a feature perspective. This review divides the recent methods into designed-feature based methods and learned-feature based methods, and summarizes the advantages, disadvantages and application scenarios. Furthermore, this paper also summarizes the performance metrics for vision-based defect recognition methods. And some challenges and development trends are also discussed.</p>

engineering, manufacturing, industrial,operations research & management science

What problem does this paper attempt to address?

This paper attempts to address the issue of how to achieve effective control of product quality through vision-based defect recognition technology in modern manufacturing systems, especially in the context of industrial intelligence. With the development of industrial big data, sensors can widely capture defect images, but how to accurately recognize these defects has become a research hotspot. Traditional manual recognition methods are inefficient and yield unstable results. Therefore, defect recognition technology based on computer vision has garnered extensive attention from both the industry and academia due to its rapid, economical, and stable characteristics. However, there are still many challenges in this field, such as defect feature extraction and improvement of recognition accuracy, especially in the application of emerging deep learning technologies. Therefore, this review paper aims to comprehensively review the progress of vision-based defect recognition technology in recent years, particularly from the perspective of feature extraction, to promote further development in this field.