| With the development of technology, welded steel-pipes have been widely used invarious fields of industrial production. The detecting of weld quality has also become a keyfactor in determining products’ quality. In our country, the X-ray detection technology basedon enhancer develops earlier in the field of weld ray detection. But it is mostly for spiral weld.Until recent years, straight weld and flat-panel detectors have just developed in theapplication of ray detection.According to the actual needs of the production, I develop the real-time imagingsoftware which can detect longitudinal welded pipe based on X-ray. The software can becontrolled by PLC. The flat-panel detector of this system is XRD0822produced byPerkinElmer. At the same time, this system also used the PLC equipments of SiemensCompany. This system is applied in the actual production. It is confirmed that this softwaremeets the needs of industrial production.The mechanical control part of the system can control motors through PLC and cancollect datas automatically. In the image processing section, reducing the intrinsic noise of theflat-panel detector using correction techniques. The random noise of images can be denoisedby the multi-frame acquisition. We can improve the images contrast through gray automaticadjustment. Because the images are blurred in the dynamic sequences, we emulate dynamicnoise reduction using NL-means algorithm. Experimental results show that this algorithm cansuppress the noise of the target area and can improve clarity of the dynamic image. In the part of image segmentation, we introduce the common segmentation algorithms.For the unique gray distribution of straight seam, we use the gradient threshold algorithm tosegment seam. By analyzing the common defect images of weld such as slag and porosity, weuse the improved algorithm of Canny to extract defects’ edge. After that,we fill its inside.Experiments show that the algorithm programs simply and can extract the weld defectseffectively. |
PDF Full Text Request