Automated multi-object tracking: Applications to metal combustion under XPCI

This study introduces the XPCI Multi-object Tracker (XMOT), a tool designed for the automated analysis of X-ray Phase Contrast Imaging (XPCI) videos that capture the combustion of metal composite powders. While tailored for XPCI data, the design and methods behind XMOT are general and can be applied to many kinds of scientific imaging of dynamic processes. XMOT automates the detection of particles and the construction of their trajectories, greatly improving the efficiency of data analysis. This methodology allows for the quantification of dynamic and static particle properties and has been used to demonstrate that micro-explosions occur in both spherical and non-spherical particles. Such data is crucial for evaluating combustion mechanisms and performance. Validation demonstrates that XMOT achieves about 90 % accuracy and 74 % detection coverage in the particle detection step, and shape classification accuracy of about 70 % for spherical particles and about 85 % for non-spherical particles. By automating complex, labor-intensive processes, XMOT facilitates deeper insights into the relationships between material properties and combustion performance, paving the way for advanced material design and optimization.