Characterization of defects of pulsed thermography inspections by orthogonal polynomial decomposition

Abstract

Methods in infrared nondestructive testing such as normalized thermal contrast (NTC) and thermal signal reconstruction (TSR) are based on a pixel-by-pixel time domain analysis that ignores spatial correlation. Other techniques in the frequency domain such as pulsed phase thermography (PPT) present the same disadvantage. In this paper, we propose a method to evaluate defects in composite specimens based on image decomposition into a 2D orthogonal space. We compare NTC, TSR, PPT, and principal component thermography (PCT) with the new approach by inspecting three samples of composite anisotropic materials. Without defining a sound area our method estimates the depths of defects up to 1.2 mm in carbon and glass reinforced plastic specimens. An implementation of the proposed method in Matlab is available at https://github.com/charlielito/2DOrthogonal_Polynomial_Decomposition.