Research On The Key Technologies Of Visual Information Processing For GMAW Pipe-line Backing Welding

Since the welding process is automatic and intelligent now days, how to get the informationabout welding quality has been gained more and more attention.With the rapid development ofcomputer vision technology, video monitoring has been now widely used to control quality during thecourse of welding. However, the images of welding may have so many noises because of fume,spatter, arc and so on, so a set of accurate, effective and specific image technologies for welding arenecessary.The present situations and problems of the quality control in GMAW (Gas metal arc welding)pipe-line backing welding are introduced and discussed in the paper. Based on the advantages anddisadvantages of current techniques, some pretreatments are applied to the original molten pool imagefirstly in order to smooth the image and segment interested part approximately. Then, the conceptcomponent tree is introduced and used to extract the edge of molten pool. According to the edge,some parameters about molten pool which are used for later quality control are computed. At last, thewelding seam and welding wire are located through inflection points on the pool edge.Some key technologies of visual information processing for GMAW pipe-line backing weldingare researched and completed in the paper. The experiments show the algorithms proposed in thepaper can extract molten pool edge and measure welding seam and welding wire accurately when theimages have so many adverse factors. Thus the quality control of GMAW pipe-line backing weldingcan be realized.