改进的K-means红外图像互感器分割方法

Improved K-means method for infrared image mutual inductor segmentation

针对电力系统红外图像互感器分割,提出了一种基于双重判据K-means算法的红外图像互感器分割方法。在分割过程中,主要采用K-means算法对互感器红外图像进行颜色特征提取。将红外图像依次转换为灰度图、二值图,经过图像膨胀、图像腐蚀、删除小面积区域后,根据双重判据去除二值图中互感器底座,最终将互感器从电气红外图中分割。通过互感器分割实验,能更有效地对电力系统中的互感器进行故障检测和识别。

A method based on dual criterion K-means algorithm for infrared（IR） image mutual inductor segmentation is proposed to study on infrared image segmentation of power system. In the process of realizing the segmentation of mutual inductor from IR image,the K-means algorithm is mainly used to extract the colour feature of IR image of the mutual inductor.Then infrared image is converted to grayscale and binary image and after image expansion,image corrosion and deleting small area,according to the dual criterion,mutual inductor will be eventually removed from electric IR image. Through mutual inductor segmentation experiment,fault of the mutual inductor can be identified and detected more effectively from power system.