作者: Yue Yiming , Liang Xiaohui , Liu Chen , Liu Jie
DOI: 10.1007/978-3-642-22819-3_3
关键词: Simultaneous localization and mapping 、 Tracking system 、 Computer vision 、 Augmented reality 、 Artificial intelligence 、 Computer science 、 Camera auto-calibration 、 Robustness (computer science) 、 Structure from motion 、 Global localization 、 Rigid transformation
摘要: Real-time camera tracking in previously unknown scene is attractive to a wide spectrum of computer vision applications. In Recent years, Simultaneous Localization and Mapping (SLAM) system its varieties have shown extraordinary performance. However, the robustness these systems rapid erratic motion still limited because typically used Local scheme. To overcome this limitation, we present an efficient online algorithm using Global scheme which matches features global way through two steps: First, coarse are obtained nearest feature descriptor search. Afterwards, Game Theoretic approach exploited eliminate incorrect left correct can be estimate pose. Result shows our has significantly improved motion.