基于旋转投射校准的配电室指针仪表读取方法  

Reading Method of Pointer Meters in Distribution Room Based on Rotation Projection Calibration

作　　者：唐冬来 李玉 杨平 付世峻 黄璞 邓园园 TANG Donglai;LI Yu;YANG Ping;FU Shijun;HUANG Pu;DENG Yuanyuan(Aostar Information Technology Co.,Ltd,Chengdu 610095,China)

机构地区：[1]四川中电启明星信息技术有限公司,四川成都610095

出　　处：《微型电脑应用》2023年第8期9-12,共4页Microcomputer Applications

基　　金：四川省科技计划项目(2021GFW0021)。

摘　　要：为了解决配电室指针仪表监测中存在准确率差、采集频度低的问题,基于旋转投射校准,提出配电室指针仪表自动读取方法。为了消除巡检机器人多次拍摄配电室指针仪表图片存在拍摄角度差的影响,采用区域卷积神经网络对配电室指针仪表图片进行图像旋转投射校准;采用固定标准特征变换算法对指针仪表图像进行分割,并采用颜色变换特征提取指针仪表的刻度特征。在此基础上,通过霍夫变换提取指针仪表的指针读数;在某区域配电室进行实例应用,并将所提方法与应用支持向量机方法进行比较,其结果验证了所提方法的有效性和可行性。In order to solve the problems of poor accuracy and low acquisition frequency in the monitoring of pointer instruments in distribution room,an automatic reading method is proposed based on rotary projection calibration.In order to eliminate the influence of the angle difference when the inspection robot takes the pointer images for many times,the regional convolution neural network is used to calibrate the image rotation projection of the pointer images.The fixed standard feature transform algorithm is used to segment the pointer instrument image,and the color transform feature is used to extract the scale feature of the pointer instrument.On this basis,Hough transform is used to extract the pointer reading of pointer meter.An example is applied in a regional distribution room,and the proposed method is compared with the support vector machine method.The results verify the effectiveness and feasibility of the proposed method.

关 键 词：旋转投射校准 图像分割 指针仪表 刻度特征 指针读数 

分 类 号：TN99[电子电信—信号与信息处理]