Unsupervised Clustering Method for Superquadric Fitting
A new method for fitting superquadrics to point clouds uses unsupervised clustering analysis to handle noise and outliers. Unlike prior approaches that focus on either rigid or deformable superquadrics, this unified framework treats point cloud data as clustering members and parametric surface samples as centroids. The clustering process dynamically updates centroid locations to optimize superquadric parameters, establishing a principled link between clustering and shape fitting. The method is published on arXiv under ID 2605.16779.
Key facts
- Method fits superquadrics to point clouds with noise and outliers.
- Uses unsupervised clustering analysis.
- Unifies rigid and deformable superquadric fitting.
- Point cloud data treated as clustering members.
- Parametric surface samples treated as centroids.
- Dynamic centroid updates optimize superquadric parameters.
- Published on arXiv:2605.16779.
- Applicable to shape modeling across diverse fields.
Entities
Institutions
- arXiv