The Open Automation and Control Systems Journal

2014, 6 : 24-28
Published online 2014 March 07. DOI: 10.2174/1874444301406010024
Publisher ID: TOAUTOCJ-6-24

A New Method of Visual Recognition for Weld Appearance

Wei Xian
Engineering Technology Department, Panzhihua University, Panzhihua, Sichuan, China.

ABSTRACT

The visual recognition methods of texture feature of welding seam have the important significances to realize the automatic tracking of welding seam in the steps of subsequent welding and cover welding of multi-layer welding. Starting from the point of view of making full use of texture feature of welding seam image, the method research and algorithm design are respectively carried out on the 2 steps (image processing and position identification) of visual recognition of welding seam. In the image processing, the ±45°Sobel operators are used to carry out the edge detection. In the position identification, the projection method and the regional integration method are combined so as to determine the position of middle lines of welding seams.The result shows that the proposed method is simple and easy in principal and can, at the same time, meet the timeliness and accuracy requirement.

Keywords:

Image processing, regional integration, texture of welding seam.