基于红外图像的绝缘子串钢帽和盘面区域自动提取方法  

The Automatic Extraction Method of the Insulator String's Steel Caps and Disks Area Based on Infrared Image

作　　者：付强[1] 姚建刚[1] 李唐兵[2] 汤成艳[1] 于浩祺 胡淋波 

机构地区：[1]湖南大学电气与信息工程学院,湖南长沙410082 [2]国网江西省电力科学研究院,江西南昌330096

出　　处：《红外技术》2016年第11期969-974,979,共7页Infrared Technology

基　　金：国网江西省电力公司科技项目(52182015000P)

摘　　要：钢帽和盘面区域的自动提取,是绝缘子红外智能检测的基础。针对现有提取方法的不足,提出一种基于红外图像的绝缘子串钢帽和盘面区域自动提取方法:通过灰度化、双边滤波、OTSU二值分割,进行红外图像预处理;利用特定算法,实现绝缘子串自动定位;利用绝缘子串骨架的回归直线,进行图像角度校正;根据绝缘子串区域列宽分布,分割绝缘子串,提取绝缘子片;运用傅里叶描述子进行验证识别;分离钢帽和盘面,实现绝缘子串钢帽和盘面区域的提取。实验结果表明该方法可以准确地提取绝缘子串钢帽和盘面区域,为后期绝缘子故障智能诊断奠定基础。The extraction of the steel caps and disks area is the basis of the insulator intelligent detection by infrared technology. Aiming at the shortage of the existing extraction methods, an automatic extraction method of the insulator string＇s steel caps and disks area is proposed based on infrared images. Through gray processing, bilateral filtering, and OTSU binary segmentation, infrared image preprocessing is made; the specific algorithm is used to realize the automatic location of insulator string; the linear regression of the in- sulator string＇s skeleton is used to correct the image angle; according to the distribution of insulator string＇s columns width, insulator string is segmented and insulator areas are extracted; the Fourier Descriptor is used for insulator areas identification; the steel cap and disk are separated to realize the extraction of the steel caps and disks area. The experimental results show that the method can accurately extract the steel caps and disks area of the insulator and provide the basis for the intelligent fault diagnosis of insulator.

关 键 词：绝缘子串 红外图像 双边滤波 图像分割 线性回归 傅里叶描述子 

分 类 号：TN219[电子电信—物理电子学]