Detecting Shape of Weld Defect Image in X-Ray Film by Image Processing Applied Genetic Algorithm

摘要

Several types of non-destructive testing method are used for detecting weld defects. Because the X-ray radiographic testing method is particularly useful in inspecting the inside of weld metal, it is often used in industries. However, since skilled inspectors for X-ray radiographic testing are gradually decreasing, recently several methods to detect weld defects from films automatically have been investigated to improve the quality of detection results. However, a X-ray film involves a number of noise, and defect images show very low contrast and various shape in spite of the same kinds of defect. Moreover, boundaries between defect image and background are unclear, and it makes difficult to automate the inspection of X-ray films. If a type of detected defect image is judged by expert system or neural network which learns a rule of professional inspector, the boundaries of defect image has to be detected like recognizing by human's (inspector's) sense of vision. Therefore, in this study, new image processing applied the genetic algorithms which has been investigated on the computer science for search problem were constructed, and applied to detection of defect boundaries indetail.