摘要
为了解决变电站指针式仪表读数识别中指针区域提取困难、指针中心线定位误差大以及识别精度较差等问题,针对变电站中常见的刻度分布均匀的指针式仪表,提出了一种基于深度学习的指针式仪表自动检测与识别方法。首先,利用卷积神经网络模型检测当前视野下仪表目标的包围框位置,得到仪表目标图像;然后,利用改进有效和准确的场景文本检测器(EAST)算法对检测到的仪表目标图像进行文本检测,检测出仪表图像中的文本图像,利用设计的印刷体数字识别模型对文本图像进行识别,筛选出仪表刻度数字,得到仪表刻度数字的位置信息与数值;最后,通过仪表刻度数字的位置信息提取出仪表指针直线与仪表中心,通过识别出的数值结合角度法完成仪表读数识别。通过大量实验对所提出的指针式仪表读数检测与识别方法进行验证,实验结果表明,本文所提出的仪表识别方法的平均准确率高于98.5%,对于复杂背景下指针式仪表的自动检测与识别任务具有良好的准确性与稳定性,可满足变电站实际应用需求。
In order to solve the problems of difficult extraction of pointer region,large positioning error of pointer center line and poor recognition accuracy in the reading recognition of pointer meter of substation,for the pointer meter with uniform scale distribution in substation,an automatic detection and recognition method for pointer-based instruments based on deep learning is proposed.Firstly,the convolutional neural network model is used to detect the position of the bounding box of the instrument target in the current field of view,and the instrument target image is obtained.Then,an improved efficient and accurate scene text detector(EAST)algorithm is used to detect the detected target image of the instrument and detect the text image in the instrument image.The printed image digital recognition model is used to identify the text image,and the instrument scale number is selected to obtain the position information and numerical value of the meter scale number.Finally,the meter pointer line and the instrument center are extracted through the position information of the meter scale number,and the identification is performed.The value is combined with the angle method to complete the meter reading identification.Through a large number of experiments,the proposed pointer meter reading detection and identification method is verified.The experimental results show that the average accuracy of the instrument identification method proposed in this paper is higher than 98.5%.The proposed method has good accuracy and stability for automatic detection and recognition of pointer instruments in complex background,which can meet the practical application needs of the substation.
作者
徐发兵
吴怀宇
陈志环
喻汉
Xu Fabing;Wu Huaiyu;Chen Zhihuan;Yu Han(School of Information Science and Engineering, Wuhan University of Science and Technology, Wuhan 430081)
出处
《高技术通讯》
EI
CAS
北大核心
2019年第12期1206-1215,共10页
Chinese High Technology Letters
基金
国家自然科学基金(61573263)
湖北省科技支撑计划(2015BAA018)
国家重点研发计划(2017YFC0806503)资助项目
关键词
深度学习
指针式仪表识别
卷积神经网络
改进场景文本检测器(EAST)算法
deep learning
pointer meter reading recognition
convolutional neural network
improved efficient and accurate scene text detector(EAST)algorithm
