Automatic x-ray image characterisation for non-destructive evaluation

In this paper, firstly we introduce an automatic welding defect inspection system for X-ray image
evaluation, then, a novel image segmentation approach is proposed. In this approach, we first apply
an adaptive morphological filter (AMF) with an appropriate structuring element to remove noise and
most of the background image which is useless for defect identification. Secondly, edges were derived
from edge detection by using the Sobel operator. Morphological processing is used to make the edge
continuum and calculate the defect area. The defects segmentation method is also compared with
other methods. The performance of experimental results based on the proposed approach is
evaluated and has been found to be appropriate for automatic radiography Non-Destructive Evaluation
(NDE).