I am a Ph.D. student in Mathematics at
Beihang University.
My public research interests include computer vision, 3D point clouds,
machine learning, robust point-set registration, optimal transport, and
numerical algorithms for CNC interpolation.
This homepage collects my public academic projects, technical notes, and
contact information. I maintain open-source research code on GitHub and
write occasional technical notes on my personal blog.
2026 Released Sinkhorn-CPD, the official implementation of a Computer-Aided Design paper on robust point cloud registration.
2026 Released LMM-Registration, a MATLAB implementation for robust multi-view point-set registration.
2024 Maintained cnc_interpolation, a Python/Jupyter project for CNC interpolation, feedrate planning, look-ahead planning, and toolpath smoothing.
Research
My research interests revolve around geometric computation, 3D vision,
and machine learning. Current public projects focus on robust point cloud
registration, multi-view point-set alignment, optimal transport,
probabilistic modeling, and numerical algorithms for curve interpolation.
Preprint / Notes
Technical notes and study records
[Blog]
Selected notes on mathematics, algorithms, machine learning, and engineering practice.
Publications and Projects
Sinkhorn-CPD: Robust Point Cloud Registration via Unbalanced Entropic Optimal Transport Jin Zhang,
Mingyang Zhao,
Bing Liu,
Xin Jiang
Computer-Aided Design, 2026
[Paper]
[Code]
[Data]
Official Python implementation for robust point cloud registration using unbalanced entropic optimal transport.
Robust Multi-view Registration of Point Sets with Laplacian Mixture Model Jin Zhang,
Mingyang Zhao,
Xin Jiang,
Dong-Ming Yan
Pattern Recognition: 6th Asian Conference, ACPR 2021, 2022
[Paper]
[Code]
MATLAB implementation for robust registration of multiple partially overlapping 3D point sets.
CNC Interpolation Algorithms Jin Zhang Open-source algorithm project, 2024
[Repository]
Python/Jupyter implementation of line and parametric-curve interpolation,
feedrate profiles, look-ahead velocity planning, and toolpath smoothing.