Artifact suppression in industrial ultrasonic videos is critical for enhancing the visibility of weak structural features such as flaws. Traditional approaches such as frequency-based filtering and baseline subtraction are often limited by assumptions of perfectly separated signal components or fixed spatiotemporal alignment, which rarely hold in practice. More recent studies for artifact removal in medical ultrasound have considered eigenspace filtering that relies on the different spatiotemporal coherence between the consistent artifacts (clutter in medical imaging) and the transient signals of interest. This presentation will adapt the eigenfiltering approach to industrial NDT imaging wheel implementing a Synthetic Aperture Focus Technique (SAFT) applied to flaw imaging in railroad rails using a Rolling Search Unit (RSU). The presentation clarifies some important aspects of eigenfiltering of ultrasound videos. First, it is shown that the eigenfilter effectiveness primarily stems from the large spatiotemporal autocorrelation (rather than cross-correlation) of the pseudo-stationary artifact reflections compared to the small autocorrelation of the transient flaw reflections. In addition, to address the challenges of shifting artifact positions during a scan (a very common occurrence in practical tests), the presentation will propose a novel recursive eigenfilter with rectification that is different from traditional eigenfiltering. This recursive algorithm leverages a nonnegativity constraint consistent with the physics of ultrasound imaging, which iteratively reshapes the eigenspace to optimally suppress the spatiotemporally correlated artifact reflections while highlighting the uncorrelated flaw reflections. The algorithm offers excellent convergence. Experimental results obtained from RSU scanning of both artificial and natural rail flaws demonstrate outstanding filtering performance in the presence of strong artifacts. This filtering approach is widely applicable to many imaging applications involving a scanning setup, beyond rail inspections.

With the rapid development of high-speed railway technology, the safety of solid axles has become a critical factor in ensuring the reliable operation of trains. However, the complex structural characteristics of axles pose significant challenges to nondestructive testing. On one hand, the wheel seat area, due to the wheel interference fit, is prone to stress concentration, interface roughness, and fretting corrosion, which cause strong noise interference in ultrasonic signals. High noise levels can easily conceal defect signals, greatly limiting their detectability. On the other hand, the journal area generates stable strong echo signals after bearing assembly, and conventional defect evaluation methods struggle to distinguish structural echoes from real defect signals, resulting in a high false alarm rate and continued reliance on manual judgment during the inspection process.
To address these issues, this study proposes two intelligent defect evaluation methods. First, a B-scan defect database for the wheel seat region of solid axles is constructed, and an improved YOLOv5 deep learning algorithm is introduced. By optimizing feature extraction, integrating attention mechanisms and small-object detection layers, and employing active learning to enhance data quality, high-precision recognition of wheel seat defects is achieved. Second, for the journal region affected by bearing press-fit echoes, a machine learning-based evaluation method using manually extracted features and support vector machines (SVM) is designed to effectively distinguish structural echoes from actual defect signals.
Experimental results demonstrate that both methods exhibit excellent performance in solid axle inspection: the defect detection rate of the wheel seat region is significantly improved, and the false alarm rate in the journal region is greatly reduced, providing a feasible new technical solution for intelligent safety inspection of solid axles.

Railway infrastructure faces increasingly complex defect phenomena—including rolling contact fatigue (RCF) cracks, studs, and squats—that necessitate the evolution of non-destructive testing (NDT) approaches for timely and dependable identification. This work undertakes a comparative evaluation of three widely utilized inspection modalities: micro-computed tomography (micro-CT), ultrasonic testing employing the total focusing method (UT-TFM), and eddy current testing (ECT). A rail segment removed from operational service is analyzed as a representative case study to examine the detection performance of each technique. Initial results indicate that the three inspection methods offer complementary applications depending on the required resolution. UT‑TFM enables evaluation of sectioned rails without additional cutting, micro‑CT delivers enhanced detail for characterizing specific features, and ECT provides rapid measurements with localized depth sensitivity.