Undersampled flying spot thermography using compressive sensing

Active thermography has gained traction as a powerful tool for non-destructive characterisation of materials, from defect detection to thermal or adhesion property mapping. Often, a laser is used to heat the sample under study which then re-emits, typically with a maximum of emission intensity within the infrared spectral range (10 μm - 15 μm). The analysis of the thermal response can provide information about the local properties of the material, or identify emerging or buried defects. The flying spot technique is one such active thermography tool. However, as the number of pulses increase, acquisition times and data size surge beyond acceptable levels for industrial needs. In this study we show that compressive sensing methodologies can be applied to decrease the necessary measurement points with the flying spot technique while conserving its detection capabilities and the sensitivity of the thermal properties in heterogeneous materials. We compare our results with three compressive sensing algorithms to measurements without compression and show that on images of 32 × 32 points, even 20% of points suffice to identify the multilayered material properties of our sample and reconstruct the effusivity map with an error around 10% maximum. Our results show the impact that compressive sensing can have in accelerating non-destructive testing based on flying spot measurements.