3D ultrasonic reconstruction of contour node represented voids and cracks. (January 2021)

• Record Type: Journal Article
• Authors: Yan, Bei; Chen, Lei; Wu, ChongKe; Ma, Honglin
• Is Part Of: NDT & E international. Volume 117(2021)
• Publication Date: 2021-01
• Subjects: 3D contour node representation -- Ultrasonic ray tracing -- Simultaneous iterative reconstruct technique -- Adaptive moment estimation -- 3D imaging; Nondestructive testing -- Periodicals; Contrôle non destructif -- Périodiques; Electronic journals; 620.1127
• DOI: 10.1016/j.ndteint.2020.102382
• Languages: English
• ISSNs: 0963-8695

Abstract

Abstract: Although the 2D imaging technology was widely used, it cannot provide enough information for protecting cultural relics. Base on the homogeneous material assumption, 3D contour node is therefore proposed to represent defects of different shapes inside the cultural relics. This paper describes a novel algorithm to reconstruct 3D image of defects inside cultural relics combined with dynamic chain technology and two algorithms: Simultaneous Iterative Reconstruct Technique (SIRT) and Adaptive Moment Estimation(Adam). And the latter is used to improve the former. Meanwhile, a detection scheme is designed like this : Firstly, rough position of defects will be searched, and then precise reconstruction is made for defective areas. The algorithm proposed is verified by simulation experiments. With application of the algorithm, 3D images of defects of different shapes and quantities can be obtained and the errors of reconstruction, in most cases, are lower than 10%, even the maximum error is lower than 22%. Finally, the laboratory experiment also tests the ability of the proposed algorithm. The reconstruction is done by using conventional ultrasonic probes instead of array probes which can reduce the detection cost. Highlights: 3D imaging by using conventional ultrasonic probes instead of array probes. Rough positioning before reconstruction to improve imaging resolution. A 3D contour node representation is proposed to fit the best defect shape. We combine two algorithms: SIRT and Adam to obtain convergence performance. The dynamic chain method enables the algorithm to deal with different defect shapes.