Development of an X-Ray Microtomography Method for Full-Field Discrete Particle Tracking in Sand Specimens

Description

In this proposed project, groundbreaking interdisciplinary research spanning the fields of experimental geomechanics, artificial intelligence, and computer vision will develop novel pattern-recognition techniques based on in-situ X-ray microtomography (μCT), a non-destructive radiographic testing technique for examining the microstructures of materials. These techniques will allow tracking of the full-field movement of discrete particles within a sand specimen. The major thrust of this proposed project is to extend and adapt a series of well-developed object-recognition and image-matching techniques to realize feature detection, recognition, and matching of intact or crushed sand particles via μCT-based analysis of a sand specimen during its deformation process. These techniques will include ID-Track, spherical harmonics invariant of particle morphology, three-dimensional (3D) signature of histograms of orientation, 3D fast point feature histograms, 3D shape context, point-cloud registration, and convolutional neural networks. These image-matching techniques will be appropriately evaluated and integrated according to the particle-scale morphological and microstructural characteristics of sands with various kinds of geological origins. This will establish a powerful, efficient, and robust system for discrete particle tracking in any type of sand.This proposed project will generate important impacts in fundamental and applied research in a variety of fields, such as geotechnical engineering, construction materials, biomedical engineering, powders and grains, computer vision, and artificial intelligence, and will have concomitant high economic and societal impacts. The research impacts will include a newly established hybrid feature-aided tool for discrete particle tracking that will enable accurate, efficient, and robust computation of the full-field displacement and strain of any type of granular soils. Furthermore, this proposed project will advance frontiers of knowledge on object recognition and image matching, thereby laying the foundation for future interdisciplinary research in these areas. Economic and societal impacts will be achieved through the application of the novel tool for discrete particle tracking in geotechnical engineering activities, such as site investigation, design, and infrastructure construction and maintenance, or in the manufacturing, handling, and processing of other kinds of particulate materials in industry. The results will contribute to the development of optimized engineering activities and processes, leading to improvements in geotechnical infrastructure, construction management, hazard mitigation, and living environments in Hong Kong and worldwide.