Automatic three-dimensional reconstruction of subsurface defects by segmenting ultrasonic point cloud

Ultrasonic inspection is a widely accepted nondestructive method for subsurface defect detection of materials. In the past years, a number of signal processing techniques have been adopted to improve defect visibility. However, an automatic identification method that reconstructs the three-dimensional (3D) shapes of the defects from the ultrasonic data is still desired. In this work, a 3D minimum spanning tree (3D-MST) method is proposed, which treats the ultrasonic data as a point cloud and segment it by finding a tree whose sum of edge weights is as small as possible. The segmentation results give a clear indication of the positions, sizes, and shapes of the subsurface defects contained in the materials. It is also discussed that the segmentation results can be improved by adopting robust normalization as a pretreatment step before conducting 3D-MST. The feasibility of the proposed method is illustrated by case studies on a polymer composite and a food product. (c) 2021 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved.