Spatial-Time-Step Features Fusion Algorithm in Eddy Current Pulsed Thermography

摘要

Eddy Current Pulsed Thermograph (ECPT) has received extensive attention due to high sensitivity detection with surface and subsurface cracks. However, it has unsolved detection problem of blind defect location without any prior knowledge. This paper presents a spatial-time-step features fusion algorithm to obtain defects reconstruction of whole specimen by blind scanning. The proposed method is intended to use features extraction by independent component analysis (ICA) and features selection by skewness. Finally, the optimal feature of each step is fused to obtain defects reconstruction image by using common orthogonal basis extraction (COBE). Experiments have been conducted to validate this study and it shows that this algorithm has a great effect on feature selection.