Non-destructive Evaluation (NDE) is a field that is used to identify all kinds of structural damage in an object of interest without resulting in any permanent damage or modification to the object. This field has been intensively investigated for many years. Among several research topics in this field, the supervised defect detection methods are among the most innovative and challenging. In recent years, the deep learning field of artificial intelligence has made remarkable progress in image processing applications. Deep learning has shown its ability to overcome most of the disadvantages suffered by previous existing approaches in a great number of applications. In this paper, we propose a deep learning architecture based on infrared thermography inspection intended to automatically identify defects (including internal and invisible cracks, delamination, etc.) efficiently and accurately. We studied the proposed deep learning algorithms to achieve automatic defect detection and precise localization (subsurface defects case) from different thermal image sequences. To evaluate the efficiency and robustness of the proposed methodology, specimens containing artificial defects were selected for experimental configuration.
|