基于机器学习和模板匹配的变电站仪表自动读数方法

Abstract
Figure/Table
References
Related Citation (4)

Download: PDF (8167 KB) HTML (1 KB)
Export: BibTeX | EndNote (RIS)

Abstract An automatic reading method is proposed for various types of pointer instruments captured by fixed cameras in substations. This method consists of five stages: template making, template matching, image processing, needle recognition, and geometric reading. The geometric relationship between the scale value and the angle is determined by template making. The template matching algorithm is used to locate the position of the meter panel. The image of the meter panel is extracted. The interference of light and shadow on the needle recognition is reduced by Gaussian filter and gamma transform. In order to improve the effect of image binarization in complex environment, K-means clustering algorithm is used to obtain the dynamic binarization threshold of meter panel image. In order to adapt to the round and oval deformed dial, the line segment with variable length is used to fit the needle in the binary image of the meter panel to obtain the angle of the needle, and then the reading corresponding to the angle of the needle is calculated in combination with the corresponding relationship between the angle of the main scale and the scale value. The practical application results show that for the pointer meter in the substation, this method has good robustness to the interference factors such as illumination, shadow, occlusion, inclination and deformation, and the error is less than the minimum scale interval, which meets the engineering application requirements.

Key words： meter reading needle recognition template matching K-means clustering image binarization

Received: 07 September 2023

	Service

	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	LI Hanju

Cite this article:

LI Hanju. Automatic reading method of substation meters based on machine learning and template matching[J]. Electrical Engineering, 2024, 25(1): 61-66.

URL:

http://dqjs.cesmedia.cn/EN/Y2024/V25/I1/61

[1] 孙婷, 马磊. 巡检机器人中指针式仪表示数的自动识别方法[J]. 计算机应用, 2019, 39(1): 287-291.
[2] 段志达, 魏利胜, 刘小珲, 等. 基于Hough变换圆检测和边缘模板匹配的轴承缺陷检测与定位[J]. 安徽工程大学学报, 2020, 35(4): 60-68.
[3] 杜静, 魏鸿磊, 樊双蛟, 等. 基于HOUGH变换的指针式压力表自动识别算法[J]. 机床与液压, 2020, 48(11): 70-75.
[4] 杨应彬, 尹春丽, 刘波, 等. 基于Hough变换与特征聚类的指针轮廓识别方法[J]. 机械设计与研究, 2019, 35(3): 7-11.
[5] 姚洋, 彭道刚, 王志萍. 基于改进Canny检测与Hough变换的仪表图像识别算法[J]. 上海电力大学学报, 2020, 36(2): 183-189.
[6] 高建龙, 郭亮, 吕耀宇, 等. 改进ORB和Hough变换的指针式仪表识读方法[J]. 计算机工程与应用, 2018, 54(23): 252-258.
[7] 朱添益, 戴逢哲, 程思举, 等. 基于图像处理技术的换流站智能扫描系统[J]. 电气技术, 2020, 21(4): 25-29.
[8] 汪荣贵, 朱静, 杨万挺, 等. 基于照度分割的局部多尺度Retinex算法[J]. 电子学报, 2010, 38(5): 1181-1186.
[9] 王攀峰, 赵书俊, 王双玲, 等. 一种基于Retinex原理的DR图像增强改进算法[J]. 中国体视学与图像分析, 2020, 25(1): 57-64.
[10] 王萍, 孙振明. 多级分解的Retinex低照度图像增强算法[J]. 计算机应用研究, 2020, 37(4): 1204-1209.
[11] 李贤阳, 阳建中, 杨竣辉, 等. 基于改进的直方图均衡化与边缘保持平滑滤波的红外图像增强算法[J]. 计算机应用与软件, 2019, 36(3): 96-103.
[12] 王殿伟, 王晶, 许志杰, 等. 一种光照不均匀图像的自适应校正算法[J]. 系统工程与电子技术, 2017, 39(6): 1383-1390.
[13] 裴超, 王大磊, 杨占刚, 等. 考虑时空分布的配电网站房巡检策略[J]. 电气技术, 2023, 24(1): 86-90.
[14] 陈振祥, 林培杰, 程树英, 等. 基于K-means++和混合深度学习的光伏功率预测[J]. 电气技术, 2021, 22(9): 7-13.