基於深度學習與影像處理之儀表偵測與讀數識別

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：34

、訪客IP：18.191.46.36

姓名

馬政(Zheng Ma) 查詢紙本館藏

畢業系所

電機工程學系

論文名稱

基於深度學習與影像處理之儀表偵測與讀數識別

相關論文

★ 直接甲醇燃料電池混合供電系統之控制研究	★ 利用折射率檢測法在水耕植物之水質檢測研究
★ DSP主控之模型車自動導控系統	★ 旋轉式倒單擺動作控制之再設計
★ 高速公路上下匝道燈號之模糊控制決策	★ 模糊集合之模糊度探討
★ 雙質量彈簧連結系統運動控制性能之再改良	★ 桌上曲棍球之影像視覺系統
★ 桌上曲棍球之機器人攻防控制	★ 模型直昇機姿態控制
★ 模糊控制系統的穩定性分析及設計	★ 門禁監控即時辨識系統
★ 桌上曲棍球：人與機械手對打	★ 麻將牌辨識系統
★ 相關誤差神經網路之應用於輻射量測植被和土壤含水量	★ 三節式機器人之站立控制

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

[檢視]

[下載]

本電子論文使用權限為同意立即開放。
已達開放權限電子全文僅授權使用者為學術研究之目的，進行個人非營利性質之檢索、閱讀、列印。
請遵守中華民國著作權法之相關規定，切勿任意重製、散佈、改作、轉貼、播送，以免觸法。

摘要(中)

當前時代人工智慧發展迅速，許多傳統行業都逐漸從人工密集型尋求向人工智慧的轉型。在科技研究領域，深度學習的演算法不斷改進並在多個領域都得到應用。在目前關於儀表自動識別的研究中，使用傳統影像處理方法都會有很大的局限性，往往只能針對單一儀表，而且無法適應不同環境。本文結合深度學習與影像處理，嘗試用新方法辨識多種儀表並讀出儀表指數。
本文提出一套基於深度學習與影像處理之儀表偵測與讀數識別系統，其目的是對工廠中常見的儀表如圓形指針量表、方型指針量表，以及數字量表三種，進行自動偵測與讀數識別的工作。本文的主要工作流程如下，首先使用即時物件偵測網路YOLOv3從手機採集的圖像中定位儀表區域，使用影像前處理去除圖像的干擾資訊，便於後續圖像的處理；然後根據儀表的種類，使用HoughCircles和findContours函數，霍夫直線法和極座標法分別實現儀表表盤圓心定位，指針定位；再次使用即時物件偵測網路YOLOv3識別表盤的數字,根據其坐標找出離指針最近的兩組數字，結合指標的座標計算出數值，根據其坐標找出離指針最近的兩組數字，結合指標的座標計算出數值。
本論文使用即時物件偵測網路來迅速定位儀表位置，與傳統方法相比，能夠在不同的圖像條件下迅速準確地定位出各種類型和尺寸的表盤，具有很好的抗干擾性。整個系統適用於不同量程和不同種類的儀表識別，不需要給予任何儀表資訊，在提高了適用範圍的同時真正意義上實現了儀表的智能化讀數識別。

摘要(英)

With the rapid development of artificial intelligence, nowadays many traditional industries gradually from labor-intensive industry into artificial intelligence. In the field of scientific technology, deep learning algorithm has been improved and applied in many fields, especially in image processing and pattern recognition. Currently, traditional image processing methods have many limitations in the automatic instrument recognition and most of the proposed methods were designed for recognizing only one type of specific instrument meter. Therefore, this study proposes a novel method to recognize different instrument meters and read the scales pointed in the meters.
In this study, the main purpose is that a kind of intelligent system based on deep learning and image processing is designed to recognize common instruments such as circular meters, square meters and digital scale meters in the factory. First of all, we use object detection network YOLOv3 to locate instruments of images which are captured from mobile phone and camera. Then we use image processing technology to remove interference information of the images. Secondly, according to the type of instrument, HoughCircles and findContours functions are used to confirm the central position of the instrument, and hough line transform and polar coordinate method are used to realize the pointer position. In addition, YOLOv3 is applied again to recognize the number in order to find the closest number and the second closest number to the pointer, finally, we can calculate the value of the instrument.
This thesis uses object detection network to locate the instrument position. In comparison with the traditional methods, the proposed method can rapidly locate various types and sizes of the instrument in different images and has strong anti-interference. The system can be adapted to the different range and uneven scale of the instrument, which greatly improves the application range, and really realize the intelligent instrument recognition.

關鍵字(中)

★ 深度學習
★ 影像處理
★ 儀表分類
★ 指針提取
★ 讀數識別

關鍵字(英)

論文目次

目錄
摘要 i
Abstract ii
致謝 iv
目錄 v
圖目錄 vii
表目錄 ix
第一章緒論 1
1.1研究背景與動機 1
1.2文獻回顧 1
1.3論文目標 3
1.4論文架構 3
第二章儀表的偵測與分類 5
2.1卷積神經網絡 5
2.2 YOLOv3[28]網路介紹 6
2.3 YOLOv3儀表類別偵測網絡訓練資料 7
第三章儀表圖像的預處理 10
3.1可進行儀表偵測圖像與可進行儀表讀數圖像 10
3.2儀表圖像預處理 11
3.2.1圖像灰度化 12
3.2.2圖像平滑 13
3.2.3圖像二值化 14
3.2.4圖像的形態學處理 16
第四章指針提取與讀數識別的實現 18
4.1表盤區域定位 18
4.1.1相關演算法 18
4.1.2表盤區域定位方法 19
4.2指針提取 22
4.2.1霍夫直線法[31] 22
4.2.2投影法 22
4.3表盤數字識別 24
4.4讀數識別 26
4.5 系統流程 28
第五章實驗結果 30
5.1儀表類別偵測網絡 30
5.1.1儀表類別偵測網絡訓練過程與結果 30
5.1.2儀表類別偵測網路性能測試 32
5.2指針提取效果測試 33
5.3數字識別網路 33
5.4讀數識別測試 34
第六章結論與未來展望 36
6.1結論 36
6.2未來展望 37
參考文獻 38

參考文獻

[1] T. Selvathai, J. Shalinni, S. Ramesh and K. Radhakrishnan, "Automatic interpretation of analog dials in driver′s instrumentation panel," International Conference on Advances in Electrical, Electronics, Information, Communication and Bio-Informatics (AEEICB), Chennai, 2017, pp. 411-415.
[2] R. Sablatnig and W. G. Kropatsch, "Automatic reading of analog display instruments," International Conference on Pattern Recognition, Jerusalem, Israel, 1994, pp. 794-797 vol.1.
[3] M. Yi, Z. Yang, F. Guo and J. Liu, "A clustering-based algorithm for automatic detection of automobile dashboard," Annual Conference of the IEEE Industrial Electronics Society, Beijing, 2017, pp. 3259-3264.
[4] X. Mai, W. Li, Y. Huang and Y. Yang, "An Automatic Meter Reading Method Based on One-dimensional Measuring Curve Mapping," 2018 IEEE International Conference of Intelligent Robotic and Control Engineering (IRCE), Lanzhou, 2018, pp. 69-73.
[5] J. Wang, J. Huang and R. Cheng, "Automatic Reading System for Analog Instruments Based on Computer Vision and Inspection Robot for Power Plant," International Conference on Modelling, Identification and Control (ICMIC), Guiyang, 2018, pp. 1-6.
[6] Z. A. Jaffery and A. K. Dubey, "Architecture of Noninvasive Real Time Visual Monitoring System for Dial Type Measuring Instrument," IEEE Sensors Journal, vol. 13, no. 4, pp. 1236-1244, April 2013.
[7] J. Gao, L. Guo, Y. Lv, Q. Wu and D. Mu, "Research on Algorithm of Pointer Instrument Recognition and Reading Based on the Location of the Rotation Center of the Pointer," International Conference on Mechatronics and Automation (ICMA), Changchun, 2018, pp. 1405-1410.
[8] J. Han, E. Li , B. Tao and M. Lv, "Reading recognition method of analog measuring instruments based on improved hough transform," International Conference on Electronic Measurement & Instruments, Chengdu, 2011, pp. 337-340.
[9] F. Yuan, "A method of correcting the pointer reading of deflection pointer instrument," Chinese Automation Congress (CAC), Jinan, 2017, pp. 5517-5520.
[10] W. Khan, D. Ansell, K. Kuru and M. Amina, "Automated aircraft instrument reading using real time video analysis," International Conference on Intelligent Systems (IS), Sofia, 2016, pp. 416-420.
[11] 孙慧媛, 基于机器视觉的电力仪表自动识别方法研究, 浙江理工大学碩士論文, 2019年3月.
[12] 孙浩晏, 基于机器视觉的指针式仪表读数识别系统研究,吉林大学碩士論文, 2015年6月
[13] 刘阳, 基于数字图像处理技术的指针式仪表识别系统, 山东大学碩士論文, 2016年6月.
[14] 谢吉航, 基于机器视觉的工业仪表自动测度方法与应用, 电子科技大学碩士論文, 2018年3月.
[15] 邢浩强, 杜志岐,苏波.变电站指针式仪表检测与识别方法.仪器仪表学报, 2017, (11):2813-2821.
[16] 童伟圆,葛一粟,杨程光,金一鸣,高飞.基于机器视觉的指针式仪表示数识别方法研究.计算机测量与控制,2018,(03):162-166.
[17] 许丽,石伟,方甜.巡检机器人中的指针式仪表读数识别系统.仪器仪表学报,2017,(07):1782-1790.
[18] 王长旭, 水表读数识别系统的设计与实现,北京邮电大学碩士論文, 2018年3月.
[19] 刘添花, 基于机器视觉的机械水表读数识别系统,湖南大学碩士論文, 2018年6月.
[20] 方立, 面向电力巡检机器人的仪表示数识别,浙江大学碩士論文, 2017年6月.
[21] 王晓东,魏成保,冯海荣,杨瑾.基于SVM的变电站保护室数显仪表数字识别方法.电子测量技术,2019,42(02):92-95.
[22] 刘葵, 基于深度学习的指针式仪表示数识别,华中科技大学碩士論文, 2017年5月.
[23] 尧卫, 基于机器视觉的指针仪表读数识别技术研究,南昌航空大学碩士論文, 2018年3月.
[24] 施成燕, 基于智能机器人的仪表示数识别技术与系统研制, 浙江大学碩士論文,2017年6月.
[25] A. Krizhevsky, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks," Advances in neural information processing systems, 2012, pp. 1097-1105.
[26] M. Everingham, L. Van Gool, C. Williams, J. Winn, and A. Zisserman, "The pascal visual object classes challenge ," International Journal of Computer Vision, vol. 88, no. 2, pp. 303-338, 2010.
[27] J. R. Uijlings, K. E. Van De Sande, T. Gevers, and A. W. Smeulders, "Selective search for object recognition," International Journal of Computer Vision, vol. 104, no. 2, pp. 154-171, 2013.
[28] J. Redmon, A. Farhadi, “YOLOv3: An Incremental Improvement”, Computer Vision and Pattern Recognition, 2018, arXiv:1804.02767
[29] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, "You only look once: Unified, real-time object detection," Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 779-788.
[30] tzutalin/labelImg: LabelImg is a graphical image annotation tool and label object bounding boxes in images, June 2019.
Available: https://github.com/tzutalin/labelImg
[31] Hough transform – Wikipedia, June 2019.
Available: https://en.wikipedia.org/wiki/Hough_transform

指導教授

王文俊

審核日期

2019-8-22

推文