| 卞国龙,李勇,戚顺青,王艳举,于胜红,宋美芹.基于卷积神经网络的轮胎X光图像缺陷检测[J].轮胎工业,2019,39(4):0247-0251 |
二维码（扫一下试试看！） |
| 基于卷积神经网络的轮胎X光图像缺陷检测 |
| Defect Detection of Tire X Ray Image Based on Convolutional Neural Network |
| 投稿时间：2018-04-15 修订日期：2018-04-15 |
| DOI：10.12135/j.issn.1006-8171.2019.04.0247 |
| 中文关键词: 轮胎 图像分割 深度学习 卷积神经网络 缺陷检测 |
| 英文关键词: tire image segmentation deep learning convolutional neural network defect detection |
| 基金项目: |
| 中图分类号: |
|
| 摘要点击次数: 1083 |
| 全文下载次数: 620 |
| 中文摘要: |
| 为解决常用轮胎X射线图像缺陷检测方法难以获取准确的图像特征的问题，提出一种通过卷积神经网络获取图像特征的方法。对轮胎X射线图像进行数据增强，然后建立网络模型。训练算法获取图像缺陷特征，并用训练好的模型识别图像中的缺陷。首先将参数对应的神经元分为关键和非关键部分，然后采用局部关键点和动态学习率实现参数快速调节。试验结果表明，设计的网络模型不易过拟合，参数调节快，所需时间短，检测准确率高。 |
| 英文摘要: |
| In order to solve the problem that the common X ray image defect detection method was difficult to obtain accurate image features，a method of obtaining image features through convolutional neural network was proposed.The X ray image of tire was enhanced，then the network model was established.The training algorithm was used to obtain the image defect features，and the trained model was used to identify the defects in the image.First，the neurons corresponding to the parameters were divided into critical and non critical parts.Then，the local key points and dynamic learning rate were used to achieve quick adjustment of parameters.The experiment results showed that the designed network model was more difficult to be over fitted with faster parameter adjustment，shorter time and higher accuracy. |
|
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
