复杂背景下多个绝缘子串红外图像分割方法

刘新伯; 李  昊; 陈  强; 梁鉴如; 郭梦茹

复杂背景下多个绝缘子串红外图像分割方法

引用本文：刘新伯，李昊，陈强，梁鉴如，郭梦茹.复杂背景下多个绝缘子串红外图像分割方法[J].计算技术与自动化,2022,(2):125-130

摘要点击次数: 252

全文下载次数: 0

作者	单位
刘新伯，李昊，陈强，梁鉴如，郭梦茹	（上海工程技术大学电子电气工程学院，上海 201620）

中文摘要:绝缘子保证了地面与载流体绝缘，在电力巡检中具有重要意义。通过红外热像仪可以快速准确地确定绝缘子的运行状态。针对复杂背景下，多个绝缘子串红外图像的分割问题，提出了一种新的方法。首先，使用通道阈值完成图像的初步分割；然后，通过霍夫变换的直线检测定位横向塔架，结合定向腐蚀法和全局形态学方法去除电线和横向塔架；最后，使用SNIC超像素分割算法结果与形态学处理结果进行匹配和定位，实现红外图像内多个绝缘子串的分割。实验结果表明，本方法处理后结果与参考图像的相似度达到88%，相比其他聚类分割算法可以有效地去除复杂背景，实现绝缘子串的精准分割。

中文关键词:绝缘子图像分割霍夫变换定向腐蚀 SNIC

Infrared Image Segmentation Method of Multiple Insulator Strings in Complex Background

Abstract:Insulators ensure that the ground is insulated from the carrier fluid, which is of great significance in power inspections. The thermal imaging camera can quickly and accurately determine the operating status of the insulator. Aiming at the segmentation of infrared images of multiple insulators in a complex background, a new method is proposed. First, use the channel threshold to complete the initial segmentation of the image; then, the horizontal tower is positioned through the straight line detection of the Hough transform, and the wire and the horizontal tower are removed by combining the directional corrosion method and the global morphology method; finally, use the SNIC superpixel segmentation algorithm result and the morphological processing result to match and locate, realize the segmentation of multiple insulator strings in the infrared image. The experimental results show that the similarity between the results processed by this method and the reference image reaches 88%. Compared with other clustering segmentation algorithms, it can effectively remove complex backgrounds and achieve accurate segmentation of insulators.

keywords:insulator image segmentation Hough transform directional corrosion SNIC

查看全文 查看/发表评论 下载pdf阅读器