Automatic data processing based on the skewness statistic parameter for subsurface defect detection by active infrared thermography

Abstract

Non-destructive testing (NDT) by active infrared thermography requires of data processing techniques in order to: i. improve the thermal contrast between defective and non-defective areas, and ii. reduce the large quantity of images recorded during the inspection without losing relevant information. In this paper, the use of the skewness statistic parameter is proposed for the automatic processing of thermographic sequences obtained by active thermography. The main interest is to compress data from a 3D thermogram sequence to a unique skewness parameter image containing all the relevant information about the subsurface defects. Experimental thermographic data from three CFRP specimens with different surface shape (planar, curved and trapezoidal), containing 25 artificial defects (Teflon ® inserts) at different depths and having several sizes, have been processed using the skewness parameter. The results presented herein demonstrate the potential of this technique for automatic defect detection by active thermography. 1. Introduction Active infrared thermography applied to the non-destructive testing (NDT) of materials offers a reliable, straightforward and fast means for retrieving structural information from a specimen. The technique is based on the detection of surface temperature anomalies that appear in response to the application of a thermal pulse to the specimen surface. However, active thermography presents some limitations fundamentally due to an exponential rate of attenuation of the defect signature with its depth –– a consequence of the dependence of thermography on the heat conduction processes to convey information about internal structural anomalies. Defect enhancement techniques must be applied in order to produce a successful thermal inspection [1]. Conventional image processing techniques such as: the time-derivative approach by thermographic signal reconstruction (TSR) [2], the automatic algorithm based on differential absolute contrast (DAC) [3], and the application of the Fourier transform, known as pulsed phase thermography (PPT) [4] , have been applied to pulsed thermographic inspection showing both advantages and disadvantages. Recently, principal component analysis (PCA), based on the second order statistic of data, was introduced in thermal NDT. This technique called principal component thermography (PCT) [5], besides of showing some potentialities for defect detection, it compresses the information contained in the sequence to a fewer number of PCs, the number of components being dependent on the amount and type of defects and the experimental conditions and requiring an operator's decision. The present work describes an alternative method based on the third order statistic of data called skewness, which can be used to compress the information contained in the 3D sequence into a single image. The performance of the proposed method has been evaluated for depth characterization on three carbon fibre reinforced plastic (CFRP) specimens containing several defects, having different surface geometries, and being subjected to uniform and non-uniform excitation.