电力变压器温度表指针位置识别研究

电力变压器温度表是监测变压器运行状态的关键设备，利用图像识别技术是准确获取温度表读数的有效途径之一。阐述了对变压器温度指针位置进行数字图像处理和图像识别的实现过程，设计识别流程，分析比较了不同类型的表盘指针位置识别过程中的图像处理算法。针对在光照条件较好的情况下分割指针图像所用的直接色调匹配法存在的缺陷，以及变压器温度表具有红色和黑色表针的图像特点，提出一种新的适合较差光照条件的指针区域判定法--主色提取法。实验证明，该算法计算简单、适应性强、处理效果好，为准确判定指针位置和计算指针读数奠定了基础。

Study on the Recognition of Pointer Position of Electric Power Transformer Temperature Meter

The temperature meter of power transformer is considered as the key component to monitor the running state of it,and the image recognition technology can be used as one of the effective methods to get the reading of temperature meter.The process of digital image processing and identifying the position of the transformer temperature meter-pointer is expounded in this paper,the recognition flow is designed,then the analysis and compare of image processing algorithms in the process of identifying different types of meter-pointer are given.Be aimed at the flaw of the direct hue matching in segmenting pointer image under the good illumination condition,and the characteristic that the transformer temperature meter has the images of red and black pointers,a new method is put forward,the main color extracting,which is applicable for identifying pointer area under bad illumination condition.Experimentations have proved that this method has advantages in many aspects,such as computing simply,strong applicability and perfect processing result,and it can establish the basic for identifying pointer position and computing the pointer reading.