Gradient-based image generation for thermographic material inspection

Infrared thermography is a non-contact, cost-effective, and non-destructive technique for defect inspection. Analyzing the surface temperature behavior of an object excited by a suitably designed heat source provides information on the internal structure of the object. The thermal diffusion coefficient of the material is the main physical parameter determining the surface temperature profile. Defects are typically characterized by a different thermal diffusion coefficient than the base material, leading to changes in the heat transfer model. If defect identification from thermography analysis is possible and computationally efficient, interpreting the results often requires trained users. In this work, we propose an algorithm for active thermography data analysis that generates images enabling the detection of the position and size of internal defects. Experimental results validate the approach, showing its ability to detect blind flat-top holes of 3 mm diameter and depths of 0.5 mm and 0.8 mm in a 1 mm thick DP600 steel plate. In addition, tests of the proposed technique show promising results in highlighting embedded defects in a 3D-printed polylactic acid object, proving the algorithm efficacy for the inspection of materials with different heat diffusion coefficients. These findings highlight the robustness and practicality of the proposed method for industrial applications. • Detection of internal object structure through infrared image processing. • A gradient-based algorithm for analyzing thermographic data. • A methodology to generate images allowing for easy identification of defects. • Flexibility of proposed algorithm across different materials and defect types. • Successful defect detection on metal and 3D-printed polymer plates.