Detection and classification of weld defects in industrial radiography with use of advanced AI methods

摘要

The paper presents and shortly discusses the AI algorithms created for automated detection and classification of weld defects on the basis of radiography images. The three approaches are described: with use of fuzzy logic, with use of artificial neural networks and that basing on rough sets theory. The data evaluating the accuracies of obtained classifiers as well as chosen examples are attached.