Automatic recognition algorithm of digital instrument reading in offshore booster station based on Mask-RCNN

被引：0

作者：

Tang P. ^{[1
,2
]}

Liu Y. ^{[3
]}

Wei H. ^{[2
]}

Dong X. ^{[1
]}

Yan G. ^{[3
]}

Zhang Y. ^{[4
]}

Yuan Y. ^{[4
]}

Wang Z. ^{[3
]}

Fan Y. ^{[3
]}

Ma P. ^{[2
]}

机构：

[1] China Three Gorges Corporation, Beijing

[2] School of Intelligent Engineering, Zhengzhou Institute of Aeronautics Industry Management, Zhengzhou

[3] Luoyang Institute of Electro-Optical Equipment, Aviation Industry Corporation of China, Luoyang

[4] Three Gorges New Energy Offshore Wind Power Operation and Maintenance Jiangsu Co., Ltd, Yancheng

来源：

Hongwai yu Jiguang Gongcheng/Infrared and Laser Engineering | 2021年 / 50卷

关键词：

Digital instrument recognition; Image processing; Mask-RCNN; YOLOv3;

D O I：

10.3788/IRLA20211057

中图分类号：

学科分类号：

摘要：

The offshore booster station adopts the rail hanging robot to carry out patrol inspection, and the machine vision method is used to automatically identify the digital instrument reading instead of manual recording. An automatic recognition algorithm of digital instrument reading based on Mask-RCNN deep learning method was presented. The original images of different types of digital instruments were made into data sets, trained by deep learning algorithm, the parameters of the algorithm were optimized according to the change curve of loss function, the trained model was obtained, and then the digital instrument images were recognized and analyzed. The gray world algorithm and Hough transform were used for image preprocessing, which can effectively improve the accuracy of digital recognition. Finally, the recognition performance of YOLOv3 and Mask-RCNN deep learning algorithm was compared in the experiment. The results show that the former has higher detection speed and the latter has higher accuracy. The recognition rate of the latter is 99.52%, it meets the requirement that remote monitoring of offshore booster station requires high accuracy of digital instrument reading. © 2021, Editorial Board of Journal of Infrared and Laser Engineering. All right reserved.

引用

共 17 条

[1] Duan Huichuan, Zhang Haibo, Zhang Shuguang, Et al., Research on instrument digital recognition based on fuzzy theory, Instrument Technique and Sensor, 4, pp. 37-39, (2004)
[2] Guo Shuang, Research on automatic identification method of nixie tube digital instrument, Communications Technology, 45, 4, pp. 37-39, (2004)
[3] Lu Weina, Liu Changrong, Zheng Yucai, Et al., A digital instrument character recognition method based on template matching [J], Modern Computer, 3, pp. 70-72, (2008)
[4] Li Suping, Digital instrument recognition technology based on image processing [J], Mechanicaland Electronical Engineer-ing, 19, 6, pp. 84-86, (2013)
[5] Akula A, Singh A, Ghosh R, Et al., Target recognition in infrared imagery using convolutional neural network, Proceedings of International Conference on Computer Vision and Image Processing, (2017)
[6] Girshick R, Donahue J, Darrell T, Et al., Rich feature hierarchies for accurate object detection and semantic segmen-tation, CVPR, (2014)
[7] Redmon J, Divvala S, Girshick R, Et al., You only look once: Unified, real-time object detection, Computer Vision & Pattern Recognition, (2016)
[8] Liu W, Anguelov D, Erhan D, Et al., SSD: Single shot MultiBox detector, European Conference on Computer Vision, (2016)
[9] Lin T Y, Dollar P, Girshick R, Et al., Feature pyramid networks for object detection, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (2017)
[10] Zhuo Lei, Zhou Lv, Yang Lihong, Numeral recognition of calculator based on threading method, Software Engineering, 21, 12, pp. 1-3, (2018)

← 1 2 →