Research On X-ray On-line Quantitative Detection Technology For Welding Defect

Among weld defects NDT methods, X-ray detection is a more widely used one. Now, the on-line detection of the welding defects is still done by human interpreter. This way of detection is subjective and labor-intensive. It is difficult to quantify the size of defect. It is desirable to find an effective on-line defect quantitative detection method to assist human interpreter in evaluating the quality of weld and to make the on-line detection objective and standard. This research is based on the above-mentioned.For how to improve the online detection of defects in the quantitative and automation, we do some research as follows:1. According to X-ray imaging principle, this dissertation discusses the optimized design of the x-ray imaging system, and enables to obtain high-quality x-ray detection image. The x-ray imaging system is calibrated at the same time, and theoretical basis is provided to quantitative detection of defects.2. With a view of overcoming the disadvantages of X-ray-detected images, such as low contrast, difficulty to show the details of weld defect caused by scattering, a separate scattering algorithm based on blind parameter estimating is proposed for the restraint of the scattering in the X-ray image. Based on the assumption that image is composed of independent scattering spectrum, after the analysis of the scattering polarization, we use Blind Parameter Estimation to establish the separation scattering model of X-ray, which restrain scattering of X-ray image and improve the contrast and definition of image effectively.3. According to quantitative description of X-ray images of weld defects, classification algorithm based on support vector machine is adopted for defect segmentation. The algorithm is applicability for various flaws partition, and it is also improved accuracy compared with the traditional ones. An improved optimal search method is proposed for tracking and marking defects, adopting a method based on the growth combination to fill defects. Finally, results of defect quantitative detection show that the algorithms are effective for the segmentation, tracking and filling.