Study On The Defect Detection Of Welding X-ray Image

Pipeline is one of the five big transportation industry, keeping pace with railway, highway, aviation and water transport. However, for being Cycles Long, environmental pollution, storage inconvenience and high costs in the traditional defect detection of pipe weld, all those restrict the development of Pipeline. Contrasposing the issue, this thesis has done some research works as follows:(1) Getting the point cloud of Weld appearance outline using the tool of3D laser scanner, and realizing the reconstruction of three-dimensional feature of pipe weld using the software of Geomagic Studio. For the x-ray images of weld have low contrast, noise jamming and fuzzy boundaries, it gets better weld contour and internal defect separately through image processing using the method of mathematical morphology, image subtraction and iterative threshold segmentation. It is effective, continuous and integrate compared to traditional methods.(2) Welding identification automatically is the key problem by all scholars in this field. One among the reasons is that the welding defects is complicated, changeable, extremely irregular, and those defects are different in each weld, so it is hard to set a universal defect detection system. This thesis did a great deal of research works. After calculating the feature parameters, conducted the identification and classification of the welding defects using the tree classifier, and got a better results. Furthermore, did some primary research woks on the automatic rating of welding defects, and compiled the algorithm according industrial standard, which got a better effect.