Automated woven background removal for enhanced infrared thermographic inspection of fabric composites

Infrared thermography is a promising non-destructive testing methodology for the inspection of composite materials. While there are many accounts in literature that report an excellent performance on unidirectional composite laminates, this is not the case for woven fabric composites. Their inspection is typically characterised by the presence of an imprint of the weave pattern in the thermographic recordings, which is most-often found to be disturbing for reliable anomaly detection. In this contribution, it is proposed to evaluate the thermographic data in ????-space in order to separate the weave pattern from the defect information. To this end, an algorithm is introduced to automatically perform this decomposition. The efficiency of the proposed methodology is demonstrated for several weave patterns with different defect cases.