Virtual-source total focusing imaging of leaky Rayleigh waves on generalized cylindrical surfaces using sign coherence factor and blockwise adaptive SVD

Leaky Rayleigh wave ultrasonic inspection offers noncontact, highly sensitive detection of surface and near-surface defects, making it suitable for automated inspection of generalised cylindrical components. However, surface curvature induces beam distortion, echo-information loss, and reduced coherence, generating structural noise and imaging artefacts that degrade image quality and quantitative defect-characterisation accuracy. To address these issues, this study proposes the generalised cylindrical virtual source total focusing method (CVSTFM) as the primary imaging method. CVSTFM constructs a conformal virtual-source array from the surface profile to optimise beam incidence and reduce interfacial energy distortion. To further improve image quality, sign coherence factor (SCF) weighting exploits spatial-coherence differences to suppress artefacts and enhance defect responses, while blockwise adaptive singular value decomposition (BSVD) reduces residual noise. Finite element simulations and experimental imaging tests show that the total focusing method (TFM) and virtual source total focusing method (VSTFM) exhibit defect-focusing mismatch on cylindrical surfaces, whereas CVSTFM improves focusing. Compared with CVSTFM alone, SCF-CVSTFM-BSVD increases average signal-to-noise ratio (SNR) by 65% and reduces average array performance indicator (API) by 39%. The combined method clearly characterises obliquely arranged holes, arc-arranged holes, and cracks, supporting effective inspection of generalised cylindrical components.