Discrete Trigonometric Transform Thermography for Defect Detection in Composites

Abstract

Images obtained with Pulsed Thermography (PT) are often affected by noise and non-uniform heating. Therefore, numerous advanced signal processing methods have been proposed to solve these problems and thus improve the defects detection that are below the surface of materials. Some of these techniques are Principal Component Thermography (PCT), High-Order Statistics (HOS), Thermographic Signal Reconstruction (TSR) –and its first and second derivatives–. However, none of these methods is based on the law of conservation of energy of a signal (i.e., each image or thermogram) between space and frequency domains. With this approach, we developed an algorithm to detect defects in Carbon Fiber Reinforced Plastic (CFRP) composites. To do this we evaluated four types of Discrete Cosine Transforms(DCTs):DCT-1, DCT-2, DCT-3,DCT-4; and four types of Discrete Sine Transforms(DSTs): DST-1, DST-2, DST-3,DST-4.Comparison between results of the Contrast-to-Noise Ratio (CNR) metric shows that if the proposed algorithm uses DCT-1, then it outperforms second derivative of the TSR. Furthermore, this method is robust to noise and changes in the center and shape of non-uniform heating. After an extensive literature search, no reports on the use of these transforms in pulsed thermography were found