一种基于LBP的高压开关图像特征提取方法  

Feature Extraction Method for High-voltage Switch Images Based on LBP

作　　者：张浩哲 蔡杰 陈富国 ZHANG Hao-zhe;CAI Jie;CHEN Fu-guo(Pinggao Group Co.,Ltd.,Pingdingshan 467001 China;School of Electrical Engineering,Xi'an Jiaotong University,Xi'an 712000 China)

机构地区：[1]平高集团有限公司,河南平顶山467001 [2]西安交通大学电气学院,陕西西安712000

出　　处：《自动化技术与应用》2023年第2期45-48,166,共5页Techniques of Automation and Applications

摘　　要：针对变电站在户外复杂场景下的高压开关无法准确进行位置自动判别的问题,提出一种基于LBP(Local Binary Pattern,局部二值模式)的图像特征提取方法。该方法首先对图像进行预处理,包括灰度化、平滑处理、形态学处理、分割,得到干扰少、质量高的图像。然后对该图像利用LBP算法进行特征提取,得到该区域的纹理特征向量,再与标准图像的特征向量进行拟合对比,通过MATLAB平台仿真显示该方法效果显著,为采用图像处理方法进行高压开关位置状态判别奠定了基础。A LBP-based image feature extraction method is proposed for the problem that the high-voltage switches in the substation in the complex outdoor scenes cannot be accurately and automatically discriminated in terms of position. The method first preprocesses the image, including graying, smoothing, morphological processing, and segmentation, to obtain an image with less interference and high quality. Then the image is feature extracted using LBP algorithm to get the texture feature vector of the region, and then it is fitted and compared with the feature vector of the standard image. The simulation through MATLAB platform shows that the method is effective, which lays the foundation for using image processing method to discriminate the high voltage switch position state.

关 键 词：高压开关 形态学处理 图像分割 LBP 纹理特征向量 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP277[自动化与计算机技术—计算机科学与技术]