基于特征提取的电气设备红外图像识别算法  

Infrared image recognition algorithm based on feature extraction for electrical equipment

作　　者：党晓婧 刘顺桂 朱光南 栾孟杰[4] 沈民强 DANG Xiao-jing;LIU Shun-gui;ZHU Guang-nan;LUAN Meng-jie;SHEN Min-qiang(Shenzhen Electric Power Science Research Institute,China Southern Power Grid Co.Ltd.,Shenzhen 518000,China;a.Technical Service Department,Zhejiang Dali Technology Co.Ltd.,Hangzhou 310000,China;Production Department,Zhejiang Dali Technology Co.Ltd.,Hangzhou 310000,China;School of Information Engineering,Suihua University,Suihua 152061,China)

机构地区：[1]中国南方电网有限公司深圳供电局电力科学研究院,广东深圳518000 [2]浙江大立科技股份有限公司技术服务部,杭州310000 [3]浙江大立科技股份有限公司生产部,杭州310000 [4]绥化学院信息工程学院,黑龙江缓化152061

出　　处：《沈阳工业大学学报》2023年第3期264-269,共6页Journal of Shenyang University of Technology

基　　金：国家自然科学基金项目(61033004);黑龙江省省属高等学校基本科研业务费项目(KYYWF10236190105)。

摘　　要：针对变电站设备的红外图像特征提取和识别问题,提出了基于不变矩的图像特征提取和识别算法.基于统计学中矩函数的概念,提出了用于表征图像全局特征的图像矩函数,并以二维几何矩为基础,提出了用于图像识别的6个矩不变量参数.通过对中心矩计算方法进行改进,消除了图像变换过程中的计算误差,实现了图像不变矩参数的准确计算.结果表明,所提出算法的综合识别率可达94%,能够满足实际应用需求.Aiming at the feature extraction and recognition of infrared image for substation equipment,an image feature extraction and recognition algorithm based on invariant moment was proposed.According to the concept of moment function in statistics,an image moment function was proposed to represent the global feature of images.Six moment invariant parameters based on the two-dimensional geometric moment for image recognition were proposed.Through the improvement of the calculation method for central moment,the calculation error in the process of image transformation was eliminated,and the accurate calculation of image moment invariant parameters was realized.The results show that the comprehensive recognition rate of as-proposed method can reach 94%,and can meet the needs for practical application.

关 键 词：红外图像 特征提取 矩函数 几何矩 中心矩 不变矩 图像识别 全局特征 

分 类 号：TM744[电气工程—电力系统及自动化]