Statistical post-processing approaches for active infrared thermography: a comparative study

Abstract

Active infrared thermography is one of the favourable non-destructive testing and evaluation methods popularly being used for remote inspection of various materials/products/components/structures. It captures the temperature distribution over the test material for predefined thermal stimulus onto the surface, which is further processed to detect the sub-surface anomalies/defects hidden inside the test object. Various attempts have been made by several research groups to reveal the hidden finer subsurface features with improved sensitivity and resolution. Present work highlights a principal component analysis and its extension, robust principal component analysis to inspect for sub-surface flat-bottomed hole defects inside a mild steel sample. Further, the proposed data analysis approaches and their capabilities have been compared on the temporal thermal experimental sequence for a frequency modulated incident thermal stimulus. It is clear from the obtained results that principal component analysis outperforms the robust principal component analysis in providing the information regarding the hidden defect details lying deep inside the material with enhanced signal to noise ratio leading to increased temperature contrast over the detected sub-surface defects