基于改进SSD模型的热镀锌板表面缺陷在线检测系统  

On-line detection system for surface defects of hot-dip galvanized sheet based on improved SSD model

作　　者：孙晖[1] 孙锡柯 孙燕 沈远[1] Sun Hui;Sun Xike;Sun Yan;Shen Yuan(Electrical and Electronic Experimental Teaching Demonstration Center,Zhejiang University,Hangzhou Zhejiang, 310027,China;State Grid Nanchang Electric Power Supply Company,Nanchang,330006,China)

机构地区：[1]电工电子国家级实验教学示范中心浙江大学,浙江杭州310027 [2]国网南昌供电公司,南昌330006

出　　处：《电子技术（上海）》2018年第12期107-109,106,共4页Electronic Technology

摘　　要：针对目前热镀锌板表面缺陷检测主要以人工为主,效率低,漏检率高等问题,文章提出设计实现热镀锌板表面缺陷在线检测系统。利用工业相机采集图像,调用改进SSD(SingleShotMultiBoxDetector)模型,在线检测镀锌板表面缺陷。实践表明,系统能够识别出热镀锌板孔洞镀后、原料轧伤等典型缺陷。与传统的人眼观察方式相比,本系统检测速度更快,效率更高。In view of the fact that the surface defect detection of hot-dip galvanized sheet is mainly artificial, low efficiency and high missed detection rate, the paper proposes to design an on-line detection system for surface defects of hot-dip galvanized sheet. The image was captured by an industrial camera, and the SSD(Single Shot MultiBox Detector) model was called to detect the surface defects of the galvanized sheet on-line. Practice has shown that the system can identify typical defects such as hot-dip galvanized sheet after hole plating and raw material rolling. Compared with the traditional human eye observation method, the system is faster and more efficient.

关 键 词：热镀锌板 缺陷检测 改进SSD模型 

分 类 号：TG173[金属学及工艺—金属表面处理]