Research On DE-GMAW Arc Image Processing And Matching Algorithm

The Vision sensors’ advantages of rich information, combined with constantly updated computer and image processing technology, can effectively solve large quality fluctuation in manual welding, low production efficiency, as well as the lack adaptive ability to the environment of teaching robot. This paper proposes a self-adaptive matching algorithm based on image feature according to seam initial position guiding. Meanwhile, builds DE-GMAW welding system and magnesium alloy arc sensing systems under current laboratory conditions, and develops a preliminary study based on acquired arc image.Firstly, this article introduces associated research areas of subject in detail. It contains algorithms principle and deficiencies, for instance, ABS, NC, SSDA, modified SSDA algorithm and some other feature matching algorithms. Chapter one also shows magnesium alloy welding characteristics, welding method and research status of arc image.Secondly, this paper builds DE-GMAW welding system and magnesium alloy arc sensing system. Welding system includes welding source, water-cooled system, air supply system, wire feed system and simple control system. According to the principle of DE-GMAW, chapter two designs a welding torch clamp, which can make the distance between welding torch and inclined angle easy to adjust. Sensing system includes camera, image capture card, compound filter system and computer. On the basis of AZ31B, chapter two also designs two filter schemes, which comprises narrow band optical filter, neutral density filter and heat filter. This paper also proposes a new image matching algorithm. Chapter three detailedly introduces its principle, optimization strategy and process. Besides, some major code and interfaces are explained in this chapter. Comprehensive property of this algorithm is analyzed by matching Lena image and a seam image.Finally, chapter four carries out series of welding experiments for arc images. Through making a comparison between two filter schemes under three Geometric parameters of welding torch, serial parameters are acquired. Chapter five develops image processing tests upon two methods. Based on the three specific comparative analysis, numbers of processing step, processing speed and processing effect, morphology processing is regarded as an optimum method.