Shape Modeling

Shape Modeling targets the description of the shape variation present in a population of scans with so called statistical shape models'. A major challenge in building shape models is finding correspondences between the scans. At the Vision Lab, two classes of correspondence methods are researched. On the one hand, group-wise correspondence methods based on surface parameterization, with a focus on elongated surfaces or surfaces of cylindrical topology, have been developed. On the other hand, research is done on the use of elastic surface registration to find pairwise correspondences. Next to the fundamental research, Vision Lab also focuses on applying shape models for improving fit and function of products that are worn close to or in the human body, e.g. protective gear, head sets, and implants.

Journal publications


Danckaers, F., "The Development of 3D Statistical Shape Models for Diverse Applications", Department of Physics, vol. PhD in Sciences, 2019.



Danckaers, F., T. Huysmans, M. Van Dael, P. Verboven, B. Nicolai, and J. Sijbers, "Building 3D Statistical Shape Models of Horticultural Products", Food and Bioprocess Technology, vol. 10, issue 11, pp. 2100-2112, 2017. PDF icon Download paper (1.73 MB)