| 引用本文： |
-
彭建盛,李涛涛,侯雅茹,许恒铭.基于机器学习的裂纹识别研究现状及发展趋势[J].广西科学,2021,28(3):215-228. [点击复制]
- PENG Jiansheng,LI Taotao,HOU Yaru,XU Hengming.Research Status and Development Trend of Crack Recognition Based on Machine Learning[J].Guangxi Sciences,2021,28(3):215-228. [点击复制]
|
|
| 摘要: |
| 裂纹识别一直是机器视觉领域的重要研究内容，尤其是与之相关的自动检测算法在近年来备受关注。深度学习作为机器学习的一个分支，其在裂纹识别方面已显现出强大的功能和灵活性。本文对基于机器学习的裂纹识别技术的发展情况、研究现状以及典型方法进行详细介绍：首先介绍了多种机器学习方法在裂纹识别领域的应用，并从特征提取算法和应用对象等方面介绍了支持向量机（Support Vector Machine，SVM）、K最近邻（K-Nearest Neighbor，KNN）、神经网络、决策树和随机森林等常用的分类器；其次，从网络模型、数据集和应用对象等方面介绍了深度学习方法在裂纹识别领域的应用。同时，本文还对近20年（2000-2020年）的81篇相关文献进行对比分析，认为未来金属裂纹的识别依旧是热门研究，多种算法的混合会逐渐替代单一算法成为今后的发展方向。 |
| 关键词: 裂纹识别 机器学习 特征提取 计算机视觉 图像处理 |
| DOI：10.13656/j.cnki.gxkx.20210830.004 |
| 投稿时间：2021-03-13 |
| 基金项目:广西自然科学基金项目（2018GXNSFAA281164）和广西科技大学研究生教育创新计划项目（GKYC202105）资助。 |
|
| Research Status and Development Trend of Crack Recognition Based on Machine Learning |
|
PENG Jiansheng1,2, LI Taotao1, HOU Yaru1, XU Hengming1
|
| (1.College of Electrical and Information Engineering, Guangxi University of Science and Technology, Liuzhou, Guangxi, 545000, China;2.School of Artificial Intelligence and Smart Manufacturing, Hechi University, Yizhou, Guangxi, 546300, China) |
| Abstract: |
| Crack recognition has always been an important research content in the field of machine vision, especially the related automatic detection algorithms have attracted much attention in recent years. As a branch of machine learning, deep learning has shown strong function and flexibility in crack recognition. In this paper, the development, research status and typical methods of crack recognition technology based on machine learning are introduced in detail. Firstly, the applications of various machine learning algorithms in the field of crack recognition are introduced, and the commonly used classifiers such as support vector machine, K-Nearest Neighbor, neural network, decision tree and random forest are introduced from the aspects of feature extraction algorithm and application object. Secondly, the application of deep learning in crack recognition is introduced from the aspects of network model, data set and application object. At the same time, 81 relevant literatures in recent 20 years (2000-2020) are compared and analyzed in this article. It is considered that the recognition of metal cracks is still a hot research in the future, and the mixture of multiple algorithms will gradually replace a single algorithm as the development direction in the future. |
| Key words: crack recognition machine learning feature extraction computer vision image process |
