| 153 | 2 | 55 |
| 下载次数 | 被引频次 | 阅读次数 |
基于机器视觉的螺纹测量易受到工业环境(例如灰尘、铁屑、油渍等)的干扰,且需要人工半自动干预,导致测量结果不稳定。通过加入Attention机制对R2Unet模型进行改进,提出一种基于AA R2Unet深度学习模型和隐马尔科夫模型的高精密螺纹全自动精确测量方法。首先,为了克服工业环境中灰尘、铁屑等因素的干扰,设计了AA R2Unet模型对外螺纹进行有效边缘识别与提取;然后,通过计算螺纹边缘点梯度方向特征信息,使用隐马尔可夫模型对螺纹边缘点进行分类,达到螺纹零件在测量过程中可以任意角度放置的目的。通过实际采集工件图像制作数据集进行实验验证,结果表明,基于AA R2Unet的螺纹边缘提取方法分割精度达到95.92%,基于隐马尔可夫模型的螺纹边缘点分类准确率达到86%以上,外径测量误差在0.01 mm以内。
Abstract：The thread measurement methods based on machine vision are easily disturbed by the environment(e.g. dust,iron filings, oil stains, etc.), resulting in inaccurate measurement results. This paper improves the R2 Unet model by adding the Attention mechanism, and proposes an external thread measurement method based on AA R2 Unet and hidden Markov model(HMM). Firstly, to overcome the interference of dust, iron filings et al, the AA R2 Unet model was designed to identify and extract the external threads. Secondly, the feature information on gradient direction of thread edge points is calculated, HMM was used to classify the thread edge points so that the threaded parts can be placed at any angle during the measurement. Finally, the method with the gathered dataset was evaluated. The results show that the segmentation accuracy of the thread edge extraction method based on AA R2 Unet is 95.92%, the classification accuracy of thread edge points based on HMM is above 86% and the comprehensive measurement error is within 0.01 mm.
[1] ZAWADA-TOMKIEWICZ A, WIERUCKA I. A case study in technological quality assurance of a metric screw thread[J]. Measurement, 2018, 114:208-217.
[2] PRADHAN S U, ZHANG Y Y, LI J Y, et al. Tailored granule properties using 3D printed screw geometries in twin screw granulation[J]. Powder Technology, 2019, 341:75-84.
[3] WANG W C, GUAN F N, MA S Y, et al. Measurement system of gear parameters based on machine vision[J]. Measurement and Control, 2015, 48(8):242-248.
[4]陶显，侯伟，徐德.基于深度学习的表面缺陷检测方法综述[J].自动化学报，2021, 47(5):1017-1034.
[5] FARANA R, SIOMA A, SULIGA P, et al. A method of screw thread measurement using a 3D vision system[J]. Journal of Machine Construction and Maintenance, 2018(2):7-14.
[6] GADELMAWLA E S. Computer vision algorithms for measurement and inspection of external screw threads[J]. Measurement, 2017, 100:36-49.
[7] JOHNSON D K, BOTHA T R, ELS P S. Real-time sideslip angle measurements using digital image correlation[J].Journal of Terramechanics, 2019, 81:35-42.
[8] JING M, DU Y B. Flank angle measurement based on improved Sobeloperator[J]. Manufacturing Letters, 2020, 25:44-49.
[9] LEE Y C, YEH S S. Using machine vision to develop an on-machine thread measurement system for computer numerical control lathe machines[C]//Proceedings of the International MultiConference of Engineers and Computer Scientists, March 13-15, 2019, Hong Kong, China. Hong Kong:IMECS, 2019:1-6.
[10] MIN J. Measurement method of screw thread geometric error based on machine vision[J]. Measurement and Control,2018, 51(7/8):304-310.
[11]包能胜，方海涛.连续运动螺纹尺寸自适应机器视觉检测[J].计量学报，2020, 41(9):1062-1069.
[12] BELLO I, ZOPH B, LE Q, et al. Attention augmented convolutional networks[C]//Proceedings of 2019 IEEE/CVF International Conference on Computer Vision(ICCV), October 27-November 2, 2019, Seoul, Korea(South). New York:IEEE Xplore, 2019:3285-3294.
[13] YANG C Z, FANG L C, WEI H. Learning contour-based mid-level representation for shape classification[J]. IEEE Access, 2020, 8:157587-157601.
[14] YU Z D, FENG C, LIU M Y, et al. CASENet:deep category-aware semantic edge detection[C]//Proceedings of 2017IEEE Conference on Computer Vision and Pattern Recognition(CVPR), July 21-26, 2017, Honolulu, HI, USA.New York:IEEE Xplore, 2017:1761-1770.
[15] WANG Y P, ZHAO X, LI Y, et al. Deep crisp boundaries:from boundaries to higher-level tasks[J]. IEEE Transactions on Image Processing, 2019, 28(3):1285-1298.
[16] RABI M, AMROUCH M, MAHANI Z. Cursive arabic handwriting recognition system without explicit segmentation based on hidden Markov models[J/OL]. Journal of Data Mining&Digital Humanities, 2018[2021-03-01]. https：//hal.archives-ouvertes.fr/hal-01458216.
[17] JIANG Z H, CROOKES D, GREEN B D, et al. Contextaware mouse behavior recognition using hidden Markov models[J]. IEEE Transactions on Image Processing, 2019,28(3):1133-1148.
[18] CHEN Y K, YE J B, LI J. Aggregated Wasserstein distance and state registration for hidden Markov models[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 42(9):2133-2147.
[19] RAHUL M, AGRAWAL R, KOHLI N. Layered recognition scheme for robust human facial expression recognition using modified hidden Markov model[J]. Journal of Multimedia Processing and Technologies, 2019, 10(1):18.
[20]史宝丽，周亚美，庞志峰.各向异性全变分图像去噪算法[J].南通大学学报（自然科学版），2019, 18(4):24-33.
基本信息:
DOI：10.12194/j.ntu.20210330001
中图分类号:TG85
引用信息:
[1]张堃,李子杰,瞿宏俊等.基于注意力机制和隐马尔科夫的高精密螺纹全自动精确测量[J],2021,20(03):57-66.DOI:10.12194/j.ntu.20210330001.
基金信息:
江苏省高校自然科学基金项目（18KJB510038）;; 江苏省“333工程”项目（BRA2018218）;; 国家级大学生创新创业训练计划项目（202010304065Z）