Abstract
Ensuring the durability of reinforced concrete structures is a central challenge in structural health monitoring (SHM). Corrosion of embedded steel rebars is among the most common and damaging deterioration mechanisms, yet its early detection remains difficult due to the complexity of wave propagation in concrete and the presence of strong clutter from surface layers and structural heterogeneity. Ultrasonic imaging provides a promising nondestructive tool, but conventional reconstruction approaches often suffer from limited contrast and high background interference. In this work, a novel reconstruction approach, termed Baseline Cancellation UMBIR (BACU), is proposed to effectively mitigate baseline-related artifacts while maintaining high image fidelity. The method integrates baseline subtraction into the ultrasonic model-based iterative reconstruction (UMBIR) framework, suppressing dominant direct arrivals and low-rank background responses while preserving target-related signals. This formulation improves robustness to modeling errors and limited array coverage, both of which are common in practical SHM applications. Synthetic experiments based on a concrete specimen with an embedded rebar demonstrate the advantages of the proposed approach compared to the conventional UMBIR. The proposed method achieves stronger suppression of clutter, sharper localization of the rebar, and improved preservation of contrast across varying corrosion levels. These results provide more reliable and interpretable reconstructions, advancing the use of ultrasonic imaging for nondestructive evaluation, and long-term monitoring of critical infrastructure.