Active Microwave-Thermographic Signal Reconstruction With Adaptive Polynomial Regression

Active microwave thermography (AMT) is a coupled electromagnetic (EM) and thermographic nondestructive testing and evaluation (NDT&E) technique. AMT utilizes a radiating EM source (e.g., an antenna) that induces dielectric/magnetic heating on a specimen under test (SUT). The inspection surface of the SUT is imaged with an infrared (IR) camera over the inspection time. As the thermal excitation originates from a spatially varying radiated power density, a nonuniform thermal excitation results within (or on the surface of) the SUT that is directly related to this power density. This nonuniform heating causes uncertainty in defect detection and has the potential to lead to false positives and/or negatives. To this end, thermographic signal reconstruction (TSR), a well-established thermographic signal processing technique, is considered as a method to alleviate the effects of nonuniform thermal excitation. It was found that while the application of TSR improved defect indication, a modified variant, referred to as Active Microwave TSR (or AM-TSR), further improved defect indication. AM-TSR utilizes both the heating and cooling periods within the inspection time for analysis (whereas TSR only considers the cooling period). While AM-TSR improved defect indication over traditional TSR, thermal nonuniformity within the defect indication inhibited potential quantification. Thus, an additional modification to AM-TSR is necessary to improve defect quantification potential. To this end, a conditional polynomial regression, referred to as adaptive polynomial regression (APR), is incorporated into the AM-TSR process in order to improve the representation of the thermographic signal and quantification of the indication. The entire (modified) process is referred to as AM-TSR with APR. It was found that the error associated with defect quantification using AM-TSR with APR is less than 15%. This is significant as defect quantification using other techniques (spatiotemporal variance reconstruction, thermal contrast, AM-TSR, etc.), was not possible.