Modified Thermographic Signal-to-Noise Ratio for Active Microwave Thermography

Active microwave thermography (AMT) is an active thermographic nondestructive testing and evaluation (NDT&E) technique that uses an active electromagnetic-based excitation. This excitation is achieved through a radiating antenna and is spatially nonuniform in nature. As such, the electromagnetically induced heat is also spatially nonuniform, as it is directly related to the radiated power density incident on the specimen under test (SUT). After excitation, infrared measurements on the surface of the SUT are completed using an infrared camera. Common postprocessing techniques including thermal contrast (TC) and signal-to-noise ratio (SNR) are often applied to these measured results. As these postprocessing techniques were developed for inspections with a spatially uniform thermal excitation, challenges arise when they are applied to inspections that use a nonuniform thermal excitation. To this end, this work considers two fundamental heating scenarios common in AMT: defect heating and structure heating. Defect heating occurs when the defect is the primary electromagnetic absorber in an SUT, resulting in an induced heat source at the defect location. Structure heating takes place when the surrounding structure of the SUT is the primary electromagnetic absorber (e.g., heat source), and a defect present will affect the thermal diffusion through the SUT. For each scenario, TC and SNR are calculated. The results indicate that a reformulation of SNR is required for structure heating as SNR exceeds 0 dB for cases when a defect is and is not present (and hence creates a false positive detection). As such, a new formula is proposed and implemented (SNR
$SNR_{r}$
). The new formula provides a clear indication of the presence of a defect through the calculation of variance over the cooling period (resulting in a difference of
$SNR_{r}$
variance of 9 dB2 between defect and defect-free specimens). In addition, this new definition is also successfully applied to defect heating (difference of
$SNR_{r}$
variance of 22 dB2 between cases of with and without a defect).