演讲题目：3D Shape Analysis using Riemannian Elastic Metrics
演讲人： Dr. Hamid Laga
机构：Phenomics and Bioinformatics Research Centre,University of South Australia
Summary（研究综述）.
Quantifying similarities and differences between shapes, referred to as shape analysis, is a fundamental problem and a building block to many applications. In biology, evolutionary relationships among living and extinct species are discovered through the analysis of morphological data obtained by measuring phenotypic properties of representative organisms. To understand ontogenetic development, speciation, or evolutionary adaptation, it is important to quantify the similarity or dissimilarity of objects affected or produced by the phenomena under study. In medical imaging, studying shapes of 3D anatomical structures in the brain and comparing their evolution to typical growth patterns are of particular interest because many diseases can be linked to alterations of these shapes. Shape similarity problem appears also in many other branches of science, including computer graphics, computer vision, biometrics, bioinformatics, geology, and anthropology. In these applications, one studies the shape of objects for modelling longitudinal changes (e.g., anatomical growth patterns), for modelling differences within and across populations (e.g., species evolution and interconnection), and for synthesizing shapes with given properties (e.g., 3D modelling and animation in computer graphics).
In this talk I will review the recent theoretical developments in the field of 3D shape analysis. I will particularly focus on our recent work on 3D shape analysis using elastic metrics defined on non-linear Riemannian manifolds. In this non-linear space, shapes become points and deformations become trajectories. Geodesics (i.e., shortest paths on the manifold) correspond to the optimal amount of bending and stretching needed to align one shape onto another. The length of geodesics is a proper metric that quantifies similarities between shapes. This enables us to (1) register 3D shapes even in the presence of large elastic deformations, (2) compute shape statistics (average shapes, covariance, and high order statistics) in the shape space, (3) characterize the continuous variability in shape populations with probability models such as Gaussians or Mixtures of Gaussians
I will show the application of this framework to plant phenotyping (plant shape analysis), medical data analysis, correspondence and registration between 3D shapes that undergo elastic deformations, and 3D shape symmetrisation.
The work described here was done in collaboration with Professor Anuj Srivastava (Florida State University) and Dr. Sebastian Kurtek (The Ohio State University).
Biography（个人简介）.
Hamid Laga received his M.Sc (2003) and PhD (2006) degrees in Computer Science from Tokyo Institute of Technology in the area of 3D shape analysis and retrieval. He is currently a senior research fellow at the Phenomics and Bioinformatics Research Centre (PBRC) of the University of South Australia working on 3D modelling of plants. Prior to Joining UniSA, Hamid worked as Associate Professor at the Institut Telecom, Telecom Lille1 in France (2010-2012), Assistant Professor at Tokyo Institute of Technology (2006-2010), and PostDoctoral fellow at Nara Institute of Science and Technology in Japan (2006). His research interests span various fields of computer vision, computer graphics, and pattern recognition, with a special focus on the 3D acquisition, modelling and analysis of static and deformable shapes. His contributions in these fields received the Best Paper Award at the IEEE International Conference on Shape Modeling (2006), the Best Paper Award at NICOGRAPH Paper Context (2007), the International Paper Grand Prix (Best Paper Award) by the Japan Society of Art and Science (2008), and the APRS/IAPR Best Paper Prize at the IEEE International Conference on Digital Image Computing, Techniques and Applications (DICTA) 2012.
For more details（更多详细信息）, visit: http://people.unisa.edu.au/Hamid.Laga
联系人：谢宁（ningxie@tongji.edu.cn)
