Complex-Shaped Surface Reconstruction and Internal Defect Imaging With Omnidirectional Ultrasound Virtually Synthesized by Rotating Probe

Abstract

Curved structures are quite common in many industrial fields, and are more defect-prone than the planar parts. Ultrasonic imaging of internal defects typically requires multi-degree-of-freedom robotic arms to perform surface-conformal raster scanning with focused ultrasonic probes, maintaining beam-surface orthogonality using pre-known surface contours. In this paper, a simpler inspection strategy is developed for internal defect imaging using only single-element focused ultrasonic probe. The ultrasonic signals rotationally captured at the same linear scan steps are summed to synthesize a virtual ultrasonic beam with nearly omni-directional directivity pattern, aiming to cover specimen in much wider direction. Inverse boundary scattering transform and nearest neighbor search algorithm are introduced to reconstruct curved surface contour. An adaptative virtual source synthetic aperture focusing technique imaging method is proposed to generate a fully focused image by performing delay-and-sum post-processing of the synthesized ultrasonic signals using reconstructed surface contour without requiring prerequisite for a spatial calibration between virtual source and inspected object surface. The above strategy is verified on a plexiglass block with a 21.21-mm curvature radius wavy surface and twelve 3-mm-diameter Side Drilled Holes (SDHs) in staggered depths and spacings. The average absolute position error of surface reconstruction is lower than 0.14 mm, and all SDHs are distinctly visualized with average absolute errors below 0.18 mm in positions. This work provides a promising alternative to ultrasonic imaging of curved structures because it eliminates the need of prior knowledge of surface contour, and surface conformal scanning configuration, simply using a common single-element focused probe.