A novel and robust thermal wave signal reconstruction technique for defect detection in lock-in thermography

Abstract

In lock-in thermography, given sufficient time for periodic heating, the surface temperature will evolve periodically in a sinusoidal pattern from the transient state to the steady state. A phase image at the frequency of periodic heating can be calculated using a Fourier transform of the steady-state temperature sequence for defect detection. It has been found that the transient state surface temperature possesses superior properties, which can be utilized for defect detection. As compared to the steady state, the contrast in the transient state is 60% higher. The surface temperature can be best approximated by a hybrid polynomial model, which consists of sinusoidal and ordinary polynomial functions. A novel and robust thermal wave signal reconstruction (TWSR) technique has been derived from these properties. With this technique, the hybrid polynomial function is fitted to each pixel in the thermographic sequence and the fitted coefficients are used to reconstruct phase and background levelled images. Phase images generated in this way are less prone to noise problems and the need for using Fourier transformation is eliminated. However, better defect detection has been achieved with levelled images. Results obtained using a 3 mm thick CFRP sample show that the technique is highly repeatable and probes 43% deeper than the conventional lock-in phase image technique. The high signal-to-noise ratio in the transient state also implies the possibility of earlier defect detection. Levelled images have been found to be best at exploiting this property. It is shown that the duration of periodic heating can be reduced substantially from the times necessary for conventional steady-state lock-in imaging.