|
چکیده
|
The quality of welds is very important in some critical industries such as aeronautic manufacturing, pipelines, and civil construction. One of the most important weld inspections is industrial radiography testing. The quality of radiographic images may suffer from some level of blurriness due to poor signal-to-noise ratio, scattered X-rays, and other phenomena. Enhancing image contrast and improving defect detection may be achieved by digital image processing methods. In this study, outcomes from two thresholding algorithms are analyzed and compared. Here, to enhance the radiographic images, the dynamic wavelet transforms have been used long with two different types of thresholds, i.e., the traditional thresholding (including hard thresholding method) and the modified adaptive multi-scale products thresholding (MAMPT). The qualitative operator perception results show that two thresholding methods improved images contrasts and enhanced details visualization. Better results have been achieved by MAMPT especially around the defects regions of image. These results have been obtained by evaluation of the outputs from different algorithms. In addition, MAMPT is about 1.5 orders of magnitude quicker than the hard thresholding method making it more suitable for online weld inspection systems.
|