一种基于二维Otsu阈值法改进的区域生长红外图像分割法  

An improved region growing infrared image segmentation method based on two-dimensional Otsu threshold method

作　　者：黄和尧 李英娜[1] HUANG Heyao;LI Yingna(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650031,China;Electric Power Research Institute,Yunnan Power Grid Co.,Ltd.,Kunming 650500,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,云南昆明650031 [2]云南电网有限责任公司电力科学研究院,云南昆明650500

出　　处：《电视技术》2020年第11期63-67,共5页Video Engineering

摘　　要：针对电力设备红外图像背景复杂导致后续热故障诊断不精确的问题,提出一种结合二维Otsu阈值法和区域生长法的电力设备红外图像分割法。首先,对电力设备红外图像进行霍夫直线变换,剔除多余的接线;其次,对预处理后的图像使用二维Otsu算法,生成阈值灰度图;最后,遍历阈值灰度图,找到像素点中位数为区域生长法的初始种子点进行红外图像分割。实验验证,该方法在复杂环境和保障电力设备自身特征的情况下提升了目标设备分割精准度,可以得到精准度更高且特征完整的目标设备。Aiming at the problem of inaccurate diagnosis of subsequent thermal faults due to the complicated infrared image background of power equipment,this paper proposes a power equipment infrared image segmentation method combining two-dimensional Otsu threshold method and region growing method.Firstly,the Hough linear transformation is performed on the infrared image of the power equipment to eliminate redundant wiring.Then,the two-dimensional Otsu algorithm is used on the preprocessed image to generate a threshold grayscale image.Finally,traverse the threshold grayscale image and find the median of the pixel points as the initial seed point of the region growing method to segment the infrared image.The experimental results show that the method in this paper improves the accuracy of segmentation of the target device in the complex environment and guarantees the characteristics of the power equipment itself,and can obtain a target device with higher accuracy and complete features.

关 键 词：红外图像 二维OTSU 区域生长法 电力设备 霍夫直线变化 

分 类 号：TM933[电气工程—电力电子与电力传动]