摘要
针对当前变电站SF6气压表读数准确度低,误差大的问题,进行基于图像识别的变电站SF6气压表智能读数方法研究。利用图像识别技术初步判定表盘轮廓,采用LSD算法识别局部闭合轮廓区域内直线段,计算区域内像素点的梯度幅度值,并利用霍夫变换原理抓取气压表指针特征,确定指针回转中心,建立直角坐标系筛选所识别到的直线段特征,确定主刻度线,识别所得到的字符图像,最终得到智能读数结果。实验结果表明,与传统SF6气压表读数方法相比,将图像识别技术应用到变电站SF6气压表读数过程当中能够有效提高识别准确度。实验中得到了具体的结果,避免了传统读数方法由于读数不精准所产生的误差,证实了图像识别技术在变电站SF6气压表智能读数应用过程中的有效性。
Aiming at the problem of low definition and large error of the current substation SF6 barometer reading,the intelligent reading method of substation SF6 barometer is studied based on image recognition.Dial contours were initially determined by using image recognition technology,and LSD algorithm is used to identify local closed contour line segment.The gradient of pixels in the area of amplitude values is calculated,and the principle of hough transform is used to grab the barometer pointer characteristics to determine the pointer turning center.A rectangular coordinate system is established to recognition the straight line segment features and determine the main calibration.Recognizing the character image,finally the intelligent reading results are got.The experimental results show that compared with the traditional SF6 barometer reading method,the image recognition technology applied to the reading process of SF6 barometer in substation can effectively improve the identification clarity.The specific result obtained in the experiment,which avoids the error caused by the inaccurate reading of traditional reading methods,and proves the effectiveness of image recognition technology in the application of intelligent reading of SF6 barometer in substation.
作者
宁柏锋
董召杰
NING Baifeng;DONG Zhaojie(Shenzhen Power Supply Co.Ltd,Shenzhen 518000,China;Digital Grid Research Institute,China Southern Power Grid.,Guangzhou 518000,China)
出处
《自动化与仪器仪表》
2020年第6期48-51,56,共5页
Automation & Instrumentation
基金
中国南方电网有限责任公司科技项目(No.090000KK52170124)。
关键词
图像识别
气压表
智能读数
特征提取
Image recognition
the barometer
smart reading
feature extraction
