Augmented Reality Anywhere and Anytime
The Handheld Augmented Reality
Come to ISMAR 2011
In this project, we aim to build systems which are able to reconstruct high level geometrical models of the physical world. The high level geometrical models are polygonal models comprising faces, edges and vertices. Additionally, the models also consist of useful semantic annotations. We devide this project into two sub-projects in order to study two key aspects of structural modeling: automatic structrual model reconstruction and incorporating user input into the underline automatic reconstruction.
STRUCTURAL MODELING USING A HANDHELD DEVICE EQUIPEED WITH DEPTH SENSORS (2014)
Summary: We aim to build a new modeling system which is able to reconstruct strucrual model from indoor scenes. The structure model to be created is a polygonal model consisting of faces, edges and vertices, and also simple volumes. The geometry can, but need not be, connected. Our approach exploits real-time depth sensing from currently available handheld depth sensors to automatically estimate structures of interests. Feedback is provided instantaneously through registered graphics shown in an AR interface on the mobile device. In addition, we employ constraints driven optimization in order to improve accuracy of the final model. This enables us to achieve quantitative and qualitative structural models.
Current Project Status
At the moment, we mainly focus on developing back-end system, which automatically reconstructs strucrual model. The back-end system is based on RGB-D SLAM.
copyright (c) 2014 Graz University of Technology