以刻度線為基礎之圓形及扇形儀表板數值讀取演算法的研究

A study of scale-mark-based gauge reading algorithm for circular and arc gauges

黃科凱(Ke-Kai Huang)

指導教授 ： 張瑞益

國立臺灣大學/工學院/工程科學及海洋工程學研究所/碩士(2022年)

https://doi.org/10.6342/NTU202202274

摘要

本碩士論文提出一個自動化讀取類比儀表板數值的演算法，類比儀表板種類主要分為圓形及扇形兩種，在傳統上，圓形儀表板有讀取圓形儀表板數值的方法，讀取扇形儀表板數值也有獨立的方法，本論文提出一個可以同時讀取圓形與扇形儀表板的演算法。所提演算法利用每個儀表板都擁有刻度線這個特性，使用主成分分析(principal components analysis, PCA)技術找儀表板中心位置，再擷取出主刻度線，計算每條主刻度線在儀表板中的角度，然後使用光學字元辨識(optical character recognition, OCR)相關技術來偵測儀表板中的數值，因為使用的OCR並非針對儀表板數值設計，所提演算法需處理當儀表板數值負數與含有小數點等問題，再將主刻度線與所找到的數值做配對，接著找出指針的端點並計算其相對於儀表板中心的角度，最後使用這些角度做線性內插，算出儀表板的數值。實驗結果顯示所提演算法可以正確的讀取出儀表板的數值，其平均誤差小於0.5%。

關鍵字

主成分分析 ； 光學字元辨識 ； 儀表板中心偵測 ； 儀表板數值讀取

並列摘要

Abstract
The goal of this thesis is to present an algorithm to automatically read the values of the analog circular and arc gauges. Traditionally, the circular gauges have specific methods for gauge reading. There are also independent schemes for reading the arc gauges. This thesis proposes a general algorithm for reading of circular and arc gauges. Since both the circular and arc gauges have scale marks, the proposed algorithm first utilizes principal components analysis (PCA) technology to find the first eigenvector of each scale mark. The intersection of these eigenvectors is extracted as the center of the gauge. Main scale marks are obtained and dial values are extracted by optical character recognition (OCR) technology. Since the used OCR is not designed by gauges, the proposed algorithm has to process the gauge with negative values and floating point. Main scale marks and gauge values are bound together from the viewpoint of angles. The peak of the pointer and the corresponding angles to gauge center are further extracted. Finally, the gauge value is obtained by angle interpolation. Experimental results demonstrate that the proposed algorithm can successfully read the values of a variety of circular and arc gauges. The corresponding errors are limited in 0.5%.

並列關鍵字

Principal components analysis (PCA) ； Optical character recognition (OCR) ； Gauge center detection ； Gauge value reading

參考文獻

[1] J. S. Lauridsen, J. A. Graasmé, M. Pedersen, D. G. Jensen, S. H. Andersen, and T. B. Moeslund, "Reading circular analogue gauges using digital image processing," in 14th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (Visigrapp 2019), 2019: SCITEPRESS Digital Library, pp. 373-382.

[2] T. Selvathai, S. Ramesh, and K. Radhakrishnan, "Automatic interpretation of analog dials in driver's instrumentation panel," in 2017 Third International Conference on Advances in Electrical, Electronics, Information, Communication and Bio-Informatics (AEEICB), 2017: IEEE, pp. 411-415.

[3] M. Yifan, J. Qi, W. Junjie, and T. Guohui, "An automatic reading method of pointer instruments," in 2017 Chinese Automation Congress (CAC), 2017: IEEE, pp. 1448-1453.

[4] Y. Ma, Q. J. M. S. Jiang, and Technology, "A robust and high-precision automatic reading algorithm of pointer meters based on machine vision," Measurement Science and Technology, vol. 30, no. 1, p. 015401, 2018.

[5] J. Chi, L. Liu, J. Liu, Z. Jiang, and G. J. M. P. i. E. Zhang, "Machine vision based automatic detection method of indicating values of a pointer gauge," Mathematical Problems in Engineering vol. 2015, p. 19, 2015.