无人值守表针实时视觉自动识别与分析方法

Real-time Vision Recognition and Analysis of Electric Meter Indicator under Non-attended Condition

本文提出了一种基于计算机视觉的实时表针方位自动检测方法及其在变配电无人值守中表针指示数据自动读取中的应用。在图像预处理的基础上,采用迭代阈值法对表盘图像二值化,对二值化图像进行细化处理,利用HOUGH变换求取表针的当前位置和方向,依此在子像素分辨率下实时计算表针指示数据和越限报警。该方法可用于变配电无人值守视频监控系统中。实验结果表明了该方法的有效性和好的鲁棒性。

This paper presents a computer-vision-based automatic real-time detection method of indicator direction and position of electric meters and its application in automatic reading of indicator value of electric meters used in non-attended electric power transform and distribution system. Based on image preprocessing, we made the image binary of the dial plate by using the iterative threshold method, and thinned the binary image. Then we can use the Hough Transform to acquire the present direction and position of electric meter indicator. So we can calculate indicator value under subpixel resolution and give exceed-limitation warning according this value. The experiment resuhs have illustrated that the method is effective and reliable, and the system has a good robustness.