Full-angle bolt loosening detection method based on anti-loosening lines segmentation
-
摘要: 针对现有无法基于防松线分割进行螺栓松动全角度检测的问题,文章开发了一种基于颜色分割和方向矢量运算的方法。首先,利用一种基于Lab和RGB颜色空间转换下的a分量非线性拉伸和R分量最优阈值分割方法分割防松线图像;其次,利用开运算对图像进行形态学操作;再次,通过一种角度递进最小包围方法确定防松线连通域面积最小外接矩形并确定其方向矢量;然后,基于四象限反正切函数和一定调整求解螺栓松动的全角度;最后,设计实验方案验证文中算法的可行性和精度。实验结果表明,检测算法能够实现螺栓0~360°的松动角度检测,且最大相对误差为1.43%,其精度可以满足工程实践需要,具有良好的应用前景。Abstract: In response to the existing challenge of full-angle bolt loosening detection based on anti-loosening lines segmentation, a method based on color segmentation and directional vector operations is developed. Firstly, the anti-loosening line image is segmented using a method based on the nonlinear stretching of the 'a' component under Lab and RGB color space conversion, and the optimal threshold segmentation of the 'R' component. Secondly, morphological operations are performed on the image using an opening operation. Thirdly, the direction vector of the minimum external rectangle of the anti-loosening line connected domain is determined through an angle-progressive minimum enclosing method. Then, the full-angle of bolt loosening is solved based on the four-quadrant inverse tangent function and certain adjustments. Finally, an experimental scheme is designed to verify the feasibility and accuracy of the algorithm proposed in this paper. Experimental results show that the detection algorithm can realize 0 to 360 degree bolt loosening angle detection, with a maximum relative error of 1.43%. The accuracy of the method is satisfactory for engineering practice, demonstrating promising application prospects.
-
Key words:
- bolt loosening /
- full-angle detection /
- color segmentation /
- non-linear stretching /
- vector calculation
-
表 1 相机参数
参数 参数值 视觉传感器 9 MP 光圈 f/1.6 焦距 53 mm 图像分辨率 3 024×3 024 表 2 算法选取的参数
参数 参数值 非线性拉伸参数$ {k_1} $ 4.8×10−5 非线性拉伸参数$ {k_2} $ 5 结构元B形状 方形 结构元B大小 3×3 表 3 螺栓松动全角度测量结果
实际角度/(°) 检测角度/(°) 绝对误差/(°) 相对误差/(%) 5 4.928 7 0.071 3 1.43 10 10.071 8 0.071 8 0.72 30 29.789 6 0.210 4 0.70 60 59.524 0.476 0.79 100 100.531 5 0.531 5 0.53 120 119.656 3 0.343 7 0.29 150 150.71 0.71 0.47 180 179.713 0.287 0.16 240 239.985 5 0.014 5 0.01 300 300.168 3 0.168 3 0.06 350 350.037 7 0.037 7 0.01 360 359.453 7 0.546 3 0.15 表 4 不同标线下螺栓松动全角度测量结果
实际角度/(°) 检测角度/(°) 绝对误差/(°) 相对误差/(%) 5 4.944 1 0.055 9 1.12 10 9.9506 0.049 4 0.49 30 29.941 9 0.058 1 0.19 60 59.656 8 0.343 2 0.57 100 100.145 5 0.145 5 0.15 120 120.213 5 0.213 5 0.18 150 150.426 8 0.426 8 0.28 180 180.578 7 0.578 7 0.32 240 240.031 7 0.031 7 0.01 300 300.198 0.198 0.07 350 350.125 9 0.125 9 0.04 360 359.555 8 0.444 2 0.12 -
[1] 赵罡,李瑾岳,徐茂程,等. 航空发动机关键装配技术综述与展望[J]. 航空学报,2022,43(10):475-507. [2] 金宏,殷银银,刘乐,等. 航天用超大规格GH4169高温合金螺栓热镦工艺[J]. 锻压技术,2022,47(6):55-60. doi: 10.13330/j.issn.1000-3940.2022.06.007 [3] 王国亮,易小冬,谢清程,等. 调距桨叶根螺栓空泡损伤研究[J]. 船舶工程,2021,43(12):110-114,171. doi: 10.13788/j.cnki.cbgc.2021.12.17 [4] 唐云岗. 可分离铰接式坦克越壕性能仿真研究[D]. 长沙:中南大学,2009. [5] 百度百科. 12·14广东深圳地铁踩踏事故[EB/OL]. [2022-04-19]. https://baike.baidu.com/item/12%C2%B714%E5%B9%BF%E4%B8%9C%E6%B7%B1%E5%9C%B3%E5%9C%B0%E9%93%81%E8%B8%A9%E8%B8%8F%E4%BA%8B%E6%95%85/10856649?fr=aladdin. [6] 宝安政府在线网. 关于福海街道威宏志五金制品有限公司“11·6”物体打击死亡事故的调查报告[EB/OL]. (2018-01-16) [2022-04-19]. http://www.baoan.gov.cn/xxgk/zdly/aqsc/dcbg/content/post_3371308.html. [7] 彭凌云,韩虎. 关于风电机组检修维护的要点讨论[J]. 机电产品开发与创新,2022,35(6):173-176. [8] 管春玲,周金龙,卢俊业,等. 列车检修停靠状态下底板螺栓缺陷自动检测研究[J]. 电子技术与软件工程,2021(18):81-82. [9] Park J H,Kim T H,Lee K S,et al. Novel bolt-loosening detection technique using image processing for bolt joints in steel bridges[C]. The 2015 World Congress on Advances in Structural Engineering and Mechanics(ASEM15),2015:25-29. [10] Nguyen T C,Huynh T C,Ryn J Y,et al. Bolt-loosening identification of bolt connections by vision image-based technique[J]. Proceedings of Spie,2016,9804:1-17. [11] 王勇,邓卓森,王志云,等. 基于2D/3D成像的地铁列车螺栓松动检测研究与应用[J]. 机车电传动,2021(6):147-154. doi: 10.13890/j.issn.1000-128x.2021.06.021 [12] Pan X,Yang T Y. Image-based monitoring of bolt loosening through deep-learning-based integrated detection and tracking[J]. Computer Aided Civil and Infrastructure Engineering,2021,37(10):1-16. [13] 杨培盛,王华军,张用,等. 一种基于深度学习的地铁车体螺栓松动检测方法[J]. 轨道交通装备与技术,2021(5):38-41. doi: 10.13711/j.cnki.cn32-1836/u.2021.05.014 [14] 黄德青,倪思杰,秦娜. 一种基于防松线识别的列车螺栓松动检测方法:CN113469966A[P]. 2021-10-01. [15] Yang M,Luo L,Tao M. Efficient rail repair machine based on image recognition technology[J]. Journal of Physics:Conference Series,2020,1486(4):042023. [16] Song D,Xu X,Cui X,et al. Bolt looseness detection based on Canny edge detection algorithm[J]. Concurrency and Computation:Practice and Experience,2023:e7713. [17] Yang R,Guan Z,Lu J,et al. Bolt looseness detection algorithm based on 2D color space extraction and 3D depth measurement[C]//2022 IEEE 17th Conference on Industrial Electronics and Applications (ICIEA). IEEE,2022:491-497. [18] Deng X,Liu J,Gong H,et al. Detection of loosening angle for mark bolted joints with computer vision and geometric imaging[J]. Automation in Construction,2022,142:104517. doi: 10.1016/j.autcon.2022.104517 [19] 王前选,王锐锋,李虎,等. 轨道车辆螺栓松动量与预紧力视觉检测方法研究[J]. 铁道科学与工程学报,2023,20(9):3511-3524. doi: 10.19713/j.cnki.43-1423/u.t20221887 [20] Qi Y,Li P,Xiong B,et al. A two-step computer vision-based framework for bolt loosening detection and its implementation on a smartphone application[J]. Structural Health Monitoring,2022,21(5):2048-2062. doi: 10.1177/14759217211049995 [21] 陈梦涛,余粟. 基于Lab颜色空间的自适应K均值彩色图像分割方法[J]. 软件导刊,2021,20(6):230-234. doi: 10.11907/rjdk.202099 [22] Zheng X,Lei Q,Yao R,et al. Image segmentation based on adaptive K-means algorithm[J]. EURASIP Journal on Image and Video Processing,2018(1):1-10. [23] Otsu N. A threshold selection method from gray-level histograms[J]. IEEE Transactions on Systems Man & Cybernetics,2007,9(1):62-66. [24] Rafael C. Gonzalez,Richard E W. Digital image processing[M]. 4thed . London:Pearson,2018. [25] MathWorks. atan2 - Four-quadrant inverse tangent [EB/OL]. [2022-04-19]. https://ww2.mathworks.cn/help/matlab/ref/atan2.html?s_tid=srchtitle_atan2_1.