Surface Approximation by Means of Gaussian Process Latent Variable Models and Line Element Geometry

• DOI: 10.3390/math11020380
• Publication Date: 2023-01-11T09:45:58Z
• Authors: Ivan De Boi; Carl Henrik Ek; Rudi Penne
• Keywords: Parametric surface; Line (geometry)

The close relation between spatial kinematics and line geometry has been proven to be fruitful in surface detection reconstruction. However, methods based on this approach are limited simple geometric shapes that can formulated as a linear subspace of or element space. core is principal component formulation find best-fit approximant possibly noisy impartial given an unordered set points point cloud. We expand by introducing the Gaussian process latent variable model, probabilistic non-linear non-parametric dimensionality reduction following Bayesian paradigm. This allows us structure lower dimensional space for surfaces interest. show how applied approximation unsupervised segmentation mentioned above demonstrate its benefits deviate from these. Experiments conducted synthetic real-world objects.