首页 | 本学科首页   官方微博 | 高级检索  
     


Method of weld recognition based on textural feature matching
Authors:Zou Yirong  Wang Shenghua  Du Dong  Zhang Wenzeng  Chang Baohua
Affiliation:Key Laboratory for Advanced Materials Processing Technology,Ministry of Education,Tsinghua University,Beijing,100084
Abstract:In this paper an automatic visual method of seam recognizing and seam tracking based on textural feature matching was proposed,in order to recognize the weld of multi-layer or multi-pass welding in which the weld is difficult to be recognized by conventional visual methods.This method focuses on the obvious difference of image textural feature between the weld region and the base metal region,as well as the similarity of the textural features along the welding direction.The method consists of the following steps: setting image template and choosing the edge region as ROI (region of interest),extracting the image textural feature of the template and the edge region,feature matching,and recognition of weld region.Experiment showed that the method proposed was effective for weld seam recognition in multi-layer welding.
Keywords:weld region recognition  image texture  feature matching
本文献已被 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号