基于双光相机的光伏组件缺陷自动检测  

Automatic Solar Panel Defects Detection Based on Dual Vision Camera

作　　者：杨扬 武文佳 YANG Yang;WU Wen-jia(Shanghai Boonray Intelligent Technology Co.,Ltd,Shanghai 201203;College of Arts andSciences,Shanghai Dianji University,Shanghai 201306)

机构地区：[1]上海伯镭智能科技有限公司,上海201203 [2]上海电机学院文理教学部,上海201306

出　　处：《新型工业化》2019年第5期56-60,共5页The Journal of New Industrialization

基　　金：上海市科技型中小企业技术创新基金(1801H185100);上海电机学院2018年校级重点课程建设项目(项目编号A1-0288-19-027-034-MATLAB基础与应用;A1-0224-18-009-33-MATLAB基础与应用)

摘　　要：本文针对光伏组件的缺陷检测,提出一种基于双光相机的自动缺陷检测方法。通过离线标定建立双光相机间的单应性映射矩阵,在可见光图像上使用颜色阈值实现光伏组件分割,将分割掩模映射至热红外图像上,进而在热红外图像上的光伏组件内部检测缺陷。实验表明该方法可以有效降低缺陷误报率,提高检测准确度。In this paper, a dual vision camera based solar panel defects detection method is proposed. Through offline calibration, the homography matrix between dual vision camera is calculated. Solar panels are segmented from the visible image based on color threshold. By mapping the mask image to the infrared image, defects can be detected inside solar panels. Experiments show that this method could effectively decrease false detection rate and increase the detection precision.

关 键 词：双光相机 光伏组件 缺陷检测 单应矩阵 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]