摘要
为了满足对变电站巡检机器人识读指针仪表的实时性和准确性要求,提出了一种基于一维测量线映射的指针仪表图像智能识读方法。该方法使用加速稳健特征(speeded up robust features,SURF)算法进行图像匹配,利用单应矩阵确定表盘位置并将模板图像的一维测量线映射到待测图像,通过遍历待测图像上一维测量线路径上的像素点检测指针位置计算读数,最后对机器人在某500kV变电站巡检拍摄的指针仪表图像进行了算法有效性测试。结果表明,该方法简单实用,具有很好的鲁棒性和高精度,并且能达到实时性要求,适用于巡检机器人对变电站环境下各类指针仪表的自动识读。
In order to meet requirements for real-time performance and veracity of recognition pointer meter of the substation inspection robot,this paper presents a kind of intelligent recognition method for pointer meter images based on one-dimensional measuring line mapping.This method uses the speeded up robust features(SURF)algorithm to realize image matching,makes use of homography matrix to determine position of the dial and map the one-dimensional measuring line of the template image to the testing image and detects the pointer position by traversing pixel points on the one-dimensional measurement line path.According to calculating reading,the paper has tested effectiveness of the algorithm.The results show that this method is simple and useful,and has good robustness and high precision,and can meet requirement for real-time performance.It has verified this method is applicable to the inspection robot automatically recognizing and reading various kinds of substation pointer meters.
作者
黄炎
李文胜
麦晓明
杨英仪
方院生
HUANG Yan;LI Wensheng;MAI Xiaoming;YANG Yingyi;FANG Yuansheng(Electric Power Research Institute of Guangdong Power Grid Co.,Ltd.,Guangzhou,Guangdong510080,China)
出处
《广东电力》
2018年第12期80-85,共6页
Guangdong Electric Power
基金
广东电网有限责任公司科技项目(GDKJXM20161136)
