作者：曾科1, 高潮1, 扶新1, 郭永彩1, 秦琨2, 王攀峰2
作者单位：1. 重庆大学 光电技术及系统教育部重点实验室, 重庆 400044;
2. 重庆长安工业（集团）有限责任公司计量测试中心, 重庆 400023
关键词：字符识别;图像处理;特征提取;图像分割;小数点识别
摘要：
为提高数显仪表的识别精度和速度，提出一种针对多参数数显仪表的自动识别方法，并以电焊机的电流和电压作为算法验证。对于数显表显示的字符3和7，由于这两个字符的宽度原因，传统的穿线法会导致较高的错误识别率，因此提出一种用倾斜直线代替传统的竖直直线的改进方法。由于数显仪表的字符颜色种类繁多，利用V通道（HSV色彩空间）特征解决各种颜色数显的识别，并且减少计算量、提高定位精度。通过分析字符的特征，利用字符右侧边界的高度信息快速确定小数点位。实验结果表明该算法能够以较高精度实时识别电焊机上的多参数字符和小数点。静态识别率为99%，平均识别时间7.2 ms/张，相机动态识别率为98.4%，平均识别时间为8.5 ms/张。
Automatic recognition method for the digital display instrument with multi-parameters
ZENG Ke1, GAO Chao1, FU Xin1, GUO Yongcai1, QIN Kun2, WANG Panfeng2
1. Key Laboratory of Optoelectronic Technology and Systems of Ministry of Education, Chongqing University, Chongqing 400044, China;
2. Centre of Measurement Test, Chongqing Chang'an Industry(Group) Co., Ltd., Chongqing 400023, China
Abstract: A novel recognition method for the digital display instrument with multi-parameters was proposed to improve the recognition accuracy and speed in this paper. This paper mainly suggested algorithm for recognition of the current and voltage which were presented on electric welding machine. As for character ‘3’ and ‘7’ of the studied object, the traditional crossing method leaded to false recognition because of the character width. Therefore the slanted straight line was proposed to replace the original vertical line in the paper. Because of the various types of character color of the digital display instrument, value channel (HSV color space) was extracted to solve all sorts of color digital display and reduce computational cost and improve location precision. Decimal point was detected by acquiring the height information at the right edge of the character. The experimental results show that the character and decimal point can be recognized by high recognition accuracy and high real-time performance for digital display instrument which have multi-parameters. The accuracy of the algorithm method for static test and dynamic test is 99% and 98.4% respectively; their average time is 7.2 ms and 8.5 ms per image respectively.
Keywords: character recognition;image preprocessing;feature extraction;image segmentation;decimal point recognition
2018, 44(12): 122-128 收稿日期: 2018-03-02;收到修改稿日期: 2018-04-11
基金项目: 国防工业计量研究项目（JSJL2015209B001）
作者简介: 曾科(1991-),男,云南昭通市人,硕士研究生,专业方向为测试控制技术与图像处理
参考文献
[1] 车广. 现代焊接技术发展的现状及展望[J]. 中文信息, 2014(6):7-10
[2] 张庚, 李丹, 周亮, 等. 基于交点特征提取的数字识别方法研究[J]. 电子技术应用, 2015(z1):296-299
[3] TIAN L Y, XIAO J, HU X G. System of locating and recognizing characters in complex background[C]//20105th IEEE Conference on Industrial Electronics and Applications. ICIEA, 2010.
[4] LIANG C, YANG W M, LIAO Q M. An automatic interpretation method for LCD images of digital measuring instruments[C]//20114th International Congress on Image and Signal Processing. CISP, 2011.
[5] KULKARNI P H, KUTE P D. Optical numeral recognition algorithm for seven segment display[C]//Advances in Signal Processing. IEEE, 2016.
[6] 申小阳, 唐轶峻, 姜柏军, 等. 仪表的数字字符识别系统[J]. 仪表技术与传感器, 2005(6):55-57
[7] 童文超, 舒小华, 龙永红, 等. LED显示仪表的字符识别方法[J]. 湖南工业大学学报, 2014(1):67-70
[8] HUANG Z Q, WANG C F. Reading recognition of digital display instrument based on BP neutral network[C]//2008 International Conference on Computer Science and Software Engineering. CSSE, 2008.
[9] CHEN H, WANG X, XU B. Study on digital display instrument recognition for substation based on pulse coupled neural network[C]//IEEE International Conference on Information and Automation. ICIA, 2017.
[10] 唐轶峻, 申小阳, 朱雯兰, 等. 基于BP神经网络的数显仪表数字字符识别系统[J]. 电测与仪表, 2005, 42(9):42-45
[11] GHUGARDARE R P, NAROTE S P, MUKHERJI P, et al. Optical character recognition system for seven segment display images of measuring instruments[C]//TENCON 2009 IEEE Region 10 Conference. IEEE, 2009.
[12] HUANG J, LIU X P, ZHAO Q. Complex scene text binarization based on graph cut[C]//2016 IEEE Information Technology, Networking, Electronic and Automation Control Conference. IEEE, 2016.
[13] OTSU N. A threshold selection method from gray level histogram[C]//IEEE Transactions on Systems. TSMC, 1979.
[14] YI F, CHEN Y B. A pattern-based directional element extraction method for vehicle license plate detection in complex background[C]//2011 IEEE International Conference on Signal Processing, Communications and Computing. ICSPCC, 2011.
