Dynamic Scene 3D Reconstruction
Event details
| Date | 14.10.2014 |
| Hour | 14:00 › 15:00 |
| Speaker | Prof. Kun Li, B.E. Beijing University of Posts and Telecommunications, Beijing, 2006, Master and Ph.D. from Tsinghua University, Beijing, 2011. She is currently working as an Assistant Professor at School of Computer Science and Technology, Tianjin University, China. Her research interests include image/video processing, image-based modeling, dynamic scene 3-D reconstruction, and motion capture. |
| Location | |
| Category | Conferences - Seminars |
Dynamic scene 3-D reconstruction is a hot issue in computer vision and graphics. We built a multi-camera dome system to capture multi-view videos of an actor, and reconstructed the shape and motion at each time instant. We generated depth maps by solving a proposed variational model, which is regularized by four constraints to ensure the accuracy and completeness of the reconstruction. The depths of all the views are merged together to generate a complete 3-D mesh. Finally, taking this 3-D mesh as the underlying scene representation, we apply an iterative volumetric Laplacian deformation framework with a new positional-constraint computation scheme to extract the 3D motion information from the multi-view video sequences and generate time-varying shapes. We also propose a temporal-dense dynamic 3-D reconstruction method with low frame rate cameras. In this talk, I will also introduce several other research work of mine, including non-rigid structure from motion, depth optimization, photorealistic animation and 3-D print for human.
Practical information
- Informed public
- Free
- This event is internal
Organizer
- R. Boulic
Contact
- R. Boulic