A Comparative Study of Algebraic and Backprojection Reconstruction Techniques for Concrete Ultrasonic Tomography

Ultrasonic tomography is a promising nondestructive technique for evaluating the internal state of concrete elements, which is crucial for maintaining the sustainability and safety of civil infrastructures. In civil engineering, ongoing research seeks to improve the resolution of concrete imaging by exploring new image reconstruction techniques from ultrasonic test data. While recent studies demonstrate that techniques such as the simultaneous iterative reconstruction technique (SIRT), algebraic reconstruction technique (ART), and ordinary least squares (OLS) have achieved good image quality, there has been limited comparative analysis with other less explored methods such as filtered backprojection (FBP). This study aims to fill this gap by comparing these four reconstruction techniques—SIRT, ART, OLS, and FBP—in the context of concrete velocity transmission tomography. Concrete specimens with embedded artificial defects were adopted, as well as simulated numerical specimens. The results reveal varying effectiveness among the techniques, with the accuracy of reconstructions being influenced by the number of iterations, measurement lines, and mesh parameters. These factors significantly impact the localization of internal inclusions. In addition, the study emphasizes the role of the initial guess in image reconstruction, especially when using algebraic methods, and suggests that FBP could be beneficial as an initial guess in a combined approach with other techniques. Based on the comparative analysis, appropriate parameter definitions for concrete imaging are suggested for each method.