基于Transformer的陶瓷轴承表面缺陷检测方法
Ceramic Bearing Surface Defect Detection Method Based on Transformer
安冬 1胡荣华 1王丽艳 1邵萌 1李新然 1刘则通1
作者信息
- 1. 沈阳建筑大学机械工程学院,沈阳 110168
- 折叠
摘要
针对传统机器视觉检测方法中,由于陶瓷轴承滚动体表面曲率大、对比度低,表面成像模糊导致后续缺陷检测精度低的问题,提出一种基于Transformer的超分辨率残差网络.首先,网络使用残差学习策略,通过预测模糊图像与清晰图像之间的差值,实现超分辨率任务;其次,在网络上前端插入通道注意力模块和空间注意力模块并改进L2 多头自注意力模块,以增强图像纹理、改善梯度爆炸问题;最后,针对超分辨率重建任务,提出一种两阶段训练策略优化训练过程.自建陶瓷轴承表面缺陷数据集上的大量实验结果表明,所提出网络模型在客观指标与主观评价上均优于MSESR-GAN、VSDR等超分辨率算法,重建图像SSIM为0.939,PSNR为36.51 dB.
Abstract
A Transformer based super-resolution residual network is proposed for the problem of low accu-racy of subsequent defect detection due to blurred surface imaging caused by large curvature and low con-trast of ceramic bearing roller surface in traditional machine vision inspection methods.Firstly,the network uses a residual learning strategy to achieve the super-resolution task by predicting the difference between blurred and clear images;Secondly,a channel attention module and a spatial attention module are inserted in the front end of the network and the L2 multi-head self-attention module is improved to enhance the image texture and improve the gradient explosion problem;Finally,a two-stage training strategy is proposed to op-timize the training process for the super-resolution reconstruction task.The extensive experimental results on the self-built ceramic bearing surface defect dataset show that the proposed network model outperforms su-per-resolution algorithms such as MSESRGAN and VSDR in both objective metrics and subjective evalua-tion,with a reconstructed image SSIM of 0.939 and PSNR of 36.51 dB.
关键词
Si3N4陶瓷轴承/超分辨率重建/Transformer/图像恢复/图像增强Key words
Si3N4 ceramic bearing/super-resolution reconstruction/Transformer/image restoration/imge enhancement引用本文复制引用
基金项目
国家自然科学基金面上项目(51975130)
辽宁省教育厅项目(LJKMZ20220915)
出版年
2024
NETL
NSTL
维普
万方数据