The Institute for Mathematical Behavioral Sciences Colloquium Series presents
“See Through Point Clouds in 3D and Higher”
with Hongkai Zhao, Department of Mathematics, UCI
Thursday, November 14, 2013
4:00 – 5:00 p.m.
Social Science Plaza A, Room 2112
Point clouds are the simplest and most basics forms for data representation in 3D and higher, such as, point clouds from 3D laser scanner for shape representation, images or feature vectors viewed as points embedded in very high dimensional Euclidean spaces. However, these raw data or primitive embedding are far from intrinsic and concise. For examples, the 3D points (the x,y,z coordinates) are not even invariant under rigid motion or scaling, let alone non-rigid transformation such as different poses. Also data points primitively embedded in very high dimensional space usually stay on or close to a low dimensional manifold. One of the key issues is how to find the underlying intrinsic geometric structures or features from these point clouds for representation, analysis and understanding, such as, comparison, registration, recognition/classification, dimension reduction, etc. In this talk Zhao will present recent work on developing models and computational tools to extract intrinsic geometric quantities and features for point clouds in 3D and higher with applications to shape analysis and manifold learning.
For further information, please contact Joanna Kerner, email@example.com or 949-824-8651.