Design And Implementation Of An Experimental Platform To Detect Defects In Weld Images Based On EDGE Detection

Welding is an important process in the mechanical industrialized society. In order to ensure the quality of welding, defect detection is needed. If welding defects is in the inside of the steel pipe, the way of non-destructive testing can be taken, such as X-rays. Early, X-ray image is mainly checked manually. Image processing is the theoretical basis of automated defect detection, but X-ray picture is complex and it is a challenge for image processing algorithms. At present, there are a variety of defect detection methods. These methods have used many universal image processing algorithms, which have something in common, such as noise reduction processing. In order to study the effect of the respective image processing algorithms on the welding defects chart more easily, an experimental platform can be designed and developed. The algorithms needed can be integrated in one system,, that can increase the reusability of the module function.In this paper, mainstream defect detection methods are studied. And basic image processing algorithms which are always used in those methods are implemented and integrated in an experimental platform to expand the variety of defect detection experiments and to get the applicability of various algorithms. This article mainly research content includes the following points:Firstly, the characteristics of the welding of the X-ray acquisition pictures are studied and the characteristics of various welding defects are summarized. Secondly, image processing algorithms used in the defect detection method of the current mainstream are analyzed, such as histogram stretching and edge detection algorithm. These algorithms are realized for16-bit TIFF pictures, and the corresponding user interface is designed, so that the application of these algorithms is more flexible. Thirdly, a welding defect detection image processing platform is achieved. This platform has integrated defect detection generic function in two aspects:first, the common functions of image processing, such as mathematical morphology processing; second, the essential functions in detect defects, such as determining the area of steel pipe. Fourthly, the experiment of welding defects is taken on various types of defect pictures to find out the applicability of various algorithms for the welding defects detection of X-ray pictures.The main contribution of this paper is the design and realization of a welding defect detection experiment platform which has integrated some mainstream image processing algorithms, such as edge detection. It becomes easier to make welding defect detection experiments based on a variety of image processing algorithms. The pictures of the different types of defects have taken different exploration of defect detection methods, which has increased the reusability of those algorithms. My work has made the bedding work for further study of the higher rate of accurate defect detection method.