Automated Defect Recognition and Identification in Digital Radiography

作者：Baniukiewicz P^*

来源：Journal of Nondestructive Evaluation, 2014, 33(3): 327-334.

DOI：10.1007/s10921-013-0216-6

• Classification
• Fuzzy logic
• Artificial neural networks
• Radiography

Weld quality assurance is important for the safe exploitation of many products and constructions. This paper summarizes work on an advanced system for automated radiogram analysis. The most important parts of the process of radiogram analysis such as segmentation, thresholding and defect recognition and classification are discussed. A complex classifier composed of artificial neural networks and a fuzzy logic system is proposed and discussed in detail. The proposed classifier shows better performance and flexibility than the normal neural networks classifiers.

• 出版日期2014-9