Limited projection based in-situ X-ray tomography for hard material characterization

作者：Yu Wang, Gong-Xiang Wei, Zhenguo Zhao, Xintong Liu, Jiaxing Xu, Yiyang Fu, Zonghao Xu, Min Yue, Fuli Wang, Huiqiang Liu · 发表于：Figshare · 年份：2026 · DOI：10.6084/m9.figshare.c.8515065.v1 · 研究领域：Computer science、Materials science、Computer vision、Artificial intelligence、Structural engineering、Algorithm

Mechanical loading systems are increasingly integrated with cone-beam X-ray computed tomography (CBCT) for in-situ characterization of material mechanical properties. To achieve accurate CBCT reconstruction over 360° scanning angle, the sample chambers of mechanical loading systems are typically constructed from X-ray weakly absorbing materials. However, their limited loading capacity restricts the characterization of hard materials, and their fully enclosed design prevents sample position adjustment. Therefore, we propose a structurally flexible steel support system integrated with limited-projection in-situ CBCT to address invalid projections in partial scanning angles caused by strong X-ray attenuation of steel components, thereby significantly expanding stress loading capacity and improving operational flexibility. The improved prior image constrained compressed sensing reconstruction method incorporating total variation and non-local low-rank regularization, along with image registration algorithm (NLR-PICCS) is proposed for in-situ mechanical non-destructive testing of hard alloy materials under high stress. The core parameters, including prior weight, registration error, missing angle limits, are quantitatively analyzed and optimized with the missing projection angles (90°, 120°, and 180°). Compared with other major limited-projection reconstruction methods, both simulation and experimental results demonstrated the superiority of the proposed approach, potential to pro...