基于SVM的变电站保护室数显仪表数字识别方法
DOI:
CSTR:
作者:
作者单位:

作者简介:

通讯作者:

中图分类号:

TN9

基金项目:


SVM-based digital identification method for digital display instrument of substation protection room
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    为研发一种适合电网运检用户使用的变电站保护室巡视机器人装备,提出了一种基于支持向量机(SVM)的变电站保护室数显仪表数字识别方法,用于对机器人所拍摄的数显仪表数据进行自动化识别和监测。首先对机器人所拍摄的图片进行自定义阈值分割,并寻找连通域的最小外接矩形,获取图像中的数显仪表区域;然后对获取的数显仪表区域依次执行灰度化、中值滤波、图像增强、数字分割、自适应二值化和归一化操作,实现对数显仪表区域的预处理,获取单个数字;最后,将经过小数点处理后的单个数字图像块输入构建的多元SVM分类器,实现对单个数字的识别。实验结果表明,所提出的方法能够达到96.3%的正确识别率,可用于变电站保护室小型巡视机器人的智能巡检工作。

    Abstract:

    In order to develop a substation protection room patrol robot equipment suitable for power grid users, a digital identification method based on support vector machine (SVM) for digital display instrument of substation protection room is proposed, which is used to automatically identify and monitor the digital instrument data captured by the robot. Firstly, a custom threshold segmentation is performed on the image taken by the robot, and the minimum circumscribed rectangle of the connected domain is searched for the digital display instrument region in the image. Then the gray scale, median filtering, image enhancement, digital segmentation, adaptive binarizationand normalization operations are sequentially performed on the acquired digital display instrument region, implementing preprocessing of the digital instrument area to obtain a single number. Finally, a single digital image block processed by the decimal point is input into the constructed multivariate SVM classifier to implement the identification of a single number. The experimental results show that the proposed method can achieve 96.5% correct recognition rate and can be used for intelligent inspection of small patrol robots in substation protection room.

    参考文献
    相似文献
    引证文献
引用本文

王晓东,魏成保,冯海荣,杨瑾.基于SVM的变电站保护室数显仪表数字识别方法[J].电子测量技术,2019,42(2):92-95

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2021-07-08
  • 出版日期:
文章二维码