http://vision.gel.ulaval.ca/~ibarrac/publications/INPHY2004a.pdf

Infrared image processing and data analysis

Pulsed Phase Thermography (PPT) is a NonDestructive Testing and Evaluation (NDT&E) technique based on the Fourier Transform that can be thought as being the link between Pulsed Thermography, for which data acquisition is fast and simple; and Lock-In thermography, for which depth retrieval is straightforward. A new depth inversion technique using the phase obtained by PPT is proposed. The technique relies on the thermal diffusion length equation, i.e. μ=(α /π·f), in a similar manner as in Lock-In Thermography. The inversion problem reduces to the estimation of the blind frequency, i.e. the limiting frequency at which a defect at a particular depth presents enough phase contrast to be detected on the frequency spectra. However, an additional problem arises in PPT when trying to adequately establish the temporal parameters that will produce the desired frequency response. The decaying thermal profiles such as the ones serving as input in PPT, are non-periodic, non-band-limited functions for which, adequate sampling Δt, and truncation w(t), parameters should be selected during the signal discretization process. These parameters are both function of the depth of the defect and of the thermal properties of the specimen/defect system. A four-step methodology based on the Time-Frequency Duality of the discrete Fourier Transform is proposed to interactively determine Δt and w(t). Hence, provided that thermal data used to feed the PPT algorithm is correctly sampled and truncated, the inversion solution using the phase takes the form: z=C1μ, for which typical experimental C1 values are between 1.5 and 2. Although determination of fb is not possible when working with badly sampled data, phase profiles still present a distinctive behavior that can be used for depth retrieval purposes. An apparent blind frequency f’b, can be defined as the blind frequency at a given phase threshold φd, and be used in combination with the phase delay definition for a thermal wave: φ=z /μ, and the normalized diameter, Dn=D/z, to derive an alternative expression. Depth extraction in this case requires an additional step to recover the size of the defect.

https://corpus.ulaval.ca/jspui/bitstream/20.500.11794/18116/1/23016.pdf

Quantitative subsurface defect evaluation by pulsed phase thermography: depth retrieval with the phase

A study of defect depth using neural networks in pulsed phase thermography: modelling, noise, experiments

For many years, applications of the TNDE (Thermographic NonDestructive Evaluation) technique has been limited due to the complex non-linearity nature of related inversion problems such as defect depth estimation. Artificial neural networks have recently obtained success in revealing and providing quantitative information concerning defects in TNDE. In this paper, a three dimensional thermal model for non-homogenous materials such as carbon fiber reinforced plastic (CFRP) is first given. The modeling results are compared with the analytical solution based on Duhamel's theorem. Two back propagation neural networks (NN) as defect detector and depth estimator are then presented. Finally, simulated and experimental results are presented and discussed.

Neural network based defect detection and depth estimation in TNDE

Pulsed thermography is a popular NDE technique. In this paper, a novel statistical processing method is proposed and compared in term of defect detection and characterization with two common neural network architectures (Perceptron and Kohonen). Interest is on characterization of aluminum corrosion. The statistical method and neural network architectures use temperature, phase and amplitude data with phase and amplitude data coming from the so-called pulsed phase thermography approach. The statistical method reveals interesting performance over tested neural networks, especially in the ’interference technique,’ a combined ’two-step’ approach: detection with phase and characterization with amplitude. Theory is discussed and examples of results are presented.

Defect characterization in pulsed thermography: a statistical method compared with Kohonen and Perceptron neural networks

In this paper, a complete procedure is proposed for the calibration of a focal plane array in quantitative infrared nondestructive testing (NDT). This procedure includes vignetting correction and gray level conversion into temperature. A noise analysis is also presented in the context of pulsed infrared thermography applied to NDT. In this analysis, the authors introduce a new parameter, the noise equivalent material loss (NEML). The NEML is a global figure of merit which allows the comparison of different experimental set-ups for infrared pulsed thermography. Theory, experimental validation of the proposed concepts, and comparison with a few infrared scanning thermal imagers are presented as well.

M. Prystay

Papers

Calibration procedure for focal plane array cameras and noise equivalent material loss for quantitative thermographic NDT