小径管X射线焊缝图像缺陷识别算法

肖扬; 高炜欣; 邓国浩

doi:10.12073/j.hjxb.20230228001

小径管X射线焊缝图像缺陷识别算法

肖扬^1,,
高炜欣^{1, 2, ,},
邓国浩²

1.
西安石油大学电子工程学院, 西安, 710065
2.
西安石油大学陕西省油气井测控技术重点实验室, 西安, 710065

基金项目: 陕西省自然科学基金项目( No.2020JQ-788 )；陕西省重点研发项目( No.2020GY-179 )；陕西省重点研发计划项目(2024GX-YBXM-003).

详细信息

作者简介:
肖扬，硕士；主要研究方向为焊缝缺陷无损检测；Email: xymoh981017@163.com

通讯作者:
高炜欣，博士，教授，硕士研究生导师；Email: wxgao@xsyu.edu.cn

中图分类号: TG 441.7
计量
- 文章访问数: 137
- HTML全文浏览量: 49
- PDF下载量: 35
出版历程
- 收稿日期: 2023-02-27
- 网络出版日期: 2024-01-14
- 刊出日期: 2024-02-24

Recognition algorithm of small-diameter tube X-ray welding defect image

XIAO Yang^1,,
GAO Weixin^{1, 2, ,},
DENG Guohao²

1.
School of Electronic Engineering, Xi’an Shiyou University, Xi’an, 710065, China
2.
Key Laboratory of Gas-Oil Logging Technology, Xi’an Shiyou University, Xi’an, 710065, China

摘要

摘要:
针对小径管X射线焊缝图像缺陷检测精确率低的现状，通过对图像进行特征分析并结合稀疏字典学习，提出一种基于图像分割的小径管焊缝图像缺陷检测算法. 首先，对小径管焊缝图像进行两步图像分割获得感兴趣区域；其次，提取焊缝缺陷，得到缺陷疑似局部图像；最后，提出以不同类型原子间相关性最小为目标的小径管焊缝缺陷字典矩阵数学模型并使用K-SVD算法进行求解，利用该字典矩阵实现圆形缺陷、线形缺陷和噪声的分类鉴别. 为提高系统实时性，使用并行编程对图像分割算法进行加速. 结果表明，改进后缺陷字典矩阵对圆形缺陷识别成功率为0.974，线形缺陷识别成功率为0.967，且具有较快的识别速度，实现了小径管焊缝图像缺陷的有效识别.
- 小径管 /
- 焊缝缺陷 /
- 图像分割 /
- 稀疏字典学习
Abstract:
To address the current situation of low accuracy rate of small-diameter tube welding image defect detection, by combining image feature analysis and sparse dictionary learning, a small-diameter tube welding defect detection algorithm based on image segmentation is proposed. Firstly, using two-step image segmentation way acquires the region of interest which is in small-diameter tube welding image. Secondly, the suspected defect region is obtained by extracting welding defect. Finally, we propose a mathematical model of the dictionary matrix of small-diameter tube welding defects with the objective of minimizing correlations between different types of atoms and solve it by using K-SVD algorithm. After that, the dictionary matrix is used to classify circular defects, strip defects and noise. To improve the real-time performance of the system, we use parallel programming to accelerate the image segmentation algorithm. The results show that the recognition rate of the proposed method is 0.974 for circular defects and 0.967 for strip defects, and the recognition speed is fast, which enables the effective recognition of defects in small-diameter tube welding image.
- small-diameter tube /
- welding defect /
- image segmentation /
- sparse dictionary learning

HTML全文

图 1 小径管X射线图像

Figure 1. X-ray image of small diameter tube. (a) original image; (b) elliptical image with tilt radiography; (c) overlapping image with vertical radiography

下载: 全尺寸图片幻灯片

图 2 ROI区域提取算例1

Figure 2. ROI extraction example 1. (a) original image; (b) segmented image

下载: 全尺寸图片幻灯片

图 3 ROI区域提取算例2

Figure 3. ROI extraction example 2. (a) original image; (b) segmented image

下载: 全尺寸图片幻灯片

图 4 ROI区域提取算例3

Figure 4. ROI extraction example 3. (a) original image; (b) segmented image

下载: 全尺寸图片幻灯片

图 5 ROI区域提取算例4

Figure 5. ROI extraction example 4. (a) original image; (b) segmented image

下载: 全尺寸图片幻灯片

图 6 SDR示意图

Figure 6. SDR diagram

下载: 全尺寸图片幻灯片

图 7 疑似局部图像事例

Figure 7. Example of suspected defect region. (a) SDR of round defects; (b) SDR of linear defects; (c) SDR of noise

下载: 全尺寸图片幻灯片

图 8 K-SVD算法流程图

Figure 8. The schematic diagram of K-SVD

下载: 全尺寸图片幻灯片

图 9 缺陷识别率和不同类原子互相关性

Figure 9. Defect recognition rate and the correlation of different kinds of atoms

下载: 全尺寸图片幻灯片

图 10 SDR面积对比

Figure 10. The area comparison of SDR

下载: 全尺寸图片幻灯片

图 11 SDR灰度跨度对比

Figure 11. Gray span contrast of SDR

下载: 全尺寸图片幻灯片

表 1 测试分类结果

Table 1 The results of test classification

缺陷类型	缺陷数量 m(个)			缺陷识别率 α	准确率 A	召回率 R	精确率 P	F1-score
缺陷类型	圆形缺陷	线性缺陷	噪声	缺陷识别率 α	准确率 A	召回率 R	精确率 P	F1-score
圆形	341	6	3	0.974	0.972	0.995	0.970	0.982
线形	8	232	0	0.967
噪声	6	12	132	0.880

下载: 导出CSV

表 2 算法加速效果

Table 2 Algorithm acceleration effect

算法步骤	GPU加速前耗时${{{t_1}} \mathord{\left/ {\vphantom {{{t_1}} s}} \right. } s}$	GPU加速后耗时${{{t_2}} \mathord{\left/ {\vphantom {{{t_2}} s}} \right. } s}$
ROI分割	31.44	1.12
SDR提取	4.37	—
缺陷识别	0.01	—

下载: 导出CSV

表 3 性能对比

Table 3 Performance comparison

算法模型	准确率 $ A(\%) $	检测时间 $t{\text{/s}}$
VGG16	94.87	5.17
GoogleNet	95.57	1.36
ResNet18	96.38	1.47
YoLo V3	94.68	1.15
所提算法	97.20	5.50

下载: 导出CSV

参考文献(15)

[1]	周兆逸, 张亚南, 王肖锋, 等. 基于改正二维主成分分析的焊缝表面缺陷检测[J]. 焊接学报, 2021, 42(11): 70 − 76. Zhou Zhaoyi, Zhang Yanan, Wang Xiaofeng, et al. Weld surface defect detection based on improved two dimensional principal c-omponent analysis[J]. Transactions of the China Welding Institution, 2021, 42(11): 70 − 76.
[2]	Yang L, Wang H X, Huo B Y, et al. An automatic welding defect location algorithm based on deep learning[J]. NDT & E International:Independent Nondestructive Testing and Evaluation, 2021, 120: 102435.
[3]	Sun J, Li C, Wu X J, et al. An Effective method of weld defect detection and classification based on machine vision[J]. Ieee Transactions on Industrial Informatics, 2019, 15(12): 6322 − 6333. doi: 10.1109/TII.2019.2896357
[4]	Zhang Y, You D, Gao X, et al. Welding defects detectionbased on deep learning with multiple optical sensors during disklaser welding of thick plates[J]. Journal of Manufacturing Systems, 2019, 51: 87 − 94. doi: 10.1016/j.jmsy.2019.02.004
[5]	魏博文, 高炜欣, 刘梦溪. 激光超声结合CNN的铸件缺陷检测方法[J]. 激光与红外, 2022, 52(9): 1327 − 1334. doi: 10.3969/j.issn.1001-5078.2022.09.009 Wei Bowen, Gao Weixin, Liu Mengxi. Casting defect detection method combined with laser ultrasound and convolutional neural network[J]. Laser & Infrared, 2022, 52(9): 1327 − 1334. doi: 10.3969/j.issn.1001-5078.2022.09.009
[6]	Zhang L, Zhang Y J, Dai B C, et al. Welding defect detectionbased on local image enhancement[J]. IET Image Processing, 2019, 13(13): 2647 − 2658. doi: 10.1049/iet-ipr.2018.5840
[7]	江野, 王东, 秦小阳, 等. 奥氏体异种钢薄壁小径管焊缝裂纹超声相控阵检测[J]. 焊接学报, 2018, 39(8): 119 − 122. Jiang Ye, Wang Dong, Qin Xiaoyang, et al. Ultrasonic phased array inspection on welded crack of austenitic dissimilar ateel with thinwall and small diameter tube[J]. Transactions of the China Welding Institution, 2018, 39(8): 119 − 122.
[8]	陈小明, 谭云华, 赖传理, 等. 相控阵技术在超小径管焊缝检测应用研究[J]. 锅炉技术, 2021, 52(S1): 55 − 58. Chen Xiaoming, Tan Yunhua, Lai Chuanli, et al. The reserch of phased array testing on the butt weld of tube[J]. Boiler Technology, 2021, 52(S1): 55 − 58.
[9]	Xie L, Wang T, Zhang Y. Comparative experimental study on phased array and X-ray detection of small diameter pipe weld[J]. Journal of Physics, 2021, 1885(3): 032023.
[10]	贾潇, 赵隆. 相控阵超声与射线检测对石油化工管道缺陷检测的对比分析[J]. 中国石油和化工标准与质量, 2022, 42(15): 174 − 175. Jia Xiao, Zhao Long. Comparative analysis of ultrasonic phasedarray and X-ray detection for petrochemical pipeline defect dete-ction[J]. China Petroleum and Chemical Standard and Quality, 2022, 42(15): 174 − 175.
[11]	张小琳, 刘祎, 白贇沨, 等. 基于背景估计的焊缝缺陷检测[J]. 电子测量技术, 2022, 45(14): 116 − 122. Zhang Xiaolin, Liu Yi, Bai Yunfeng, et al. Weld defect detectionbased on background estimation[J]. Electronic Measurement Technology, 2022, 45(14): 116 − 122.
[12]	高炜欣, 胡玉衡, 穆向阳, 等. 基于聚类的埋弧焊X射线焊缝图像缺陷分割算法及缺陷模型[J]. 焊接学报, 2012, 33(4): 37 − 41. Gao Weixin, Hu Yuheng, Mu Xiangyang, et al. Study on sub-arc X-ray welding image defect segmentation algorithm and defectmodel[J]. Transactions of the China Welding Institution, 2012, 33(4): 37 − 41.
[13]	孙春红, 丁广太, 方坤. 基于稀疏字典学习的羊绒与羊毛分类[J]. 纺织学报, 2022, 43(4): 28 − 32. Sun Cunhong, Ding Guangtai, Fang Kun. Cashmere and wool classification based on sparse dictionary learning[J]. Journal of Textile Research, 2022, 43(4): 28 − 32.
[14]	Hou W H, Zhang D S, Wei Y, et al. Review on computer aided weld defect detection from radiography images[J]. Applied Sciences, 2020, 10(5): 1878 − 1893. doi: 10.3390/app10051878
[15]	修延飞, 李海超, 胡广泽, 等. 一种用于穿孔塞焊焊缝特征提取的视觉识别算法[J]. 焊接学报, 2020, 41(2): 75 − 79. doi: 10.12073/j.hjxb.20190927002 Xiu Yuanfei, Li Haichao, Hu Guangze, et al. A visual recognition algorithms for seam feature extraction of perforated plug welding[J]. Transactions of the China Welding Institution, 2020, 41(2): 75 − 79. doi: 10.12073/j.hjxb.20190927002