A CAVE Based 3D Immersive Interactive City with Gesture Interface

3D city models have greatly changed the way we interact with geographic information. However, both the visualization and interaction are limited on conventional 2D displays. This paper presents a system that visualizes 3D cities and supports gesture interaction in a fully immersive Cave Automatic Virtual Environment (CAVE). The proposed system utilizes gestures to control navigation, selection, object manipulation, and menu functions. These functions form the basis for most Virtual Reality (VR) applications, including 3D city model applications such as urban planning, virtual tourism, etc. The use of user gestures provides a more natural interaction in fully immersive VR environments. We also propose the use of pattern recognition methods, specifically a Hidden Markov Model, to facilitate real time dynamic gesture recognition and demonstrate its use for virtual menu control. As a secondary contribution, we present a novel selection method for selecting multiple objects in the VR environment. An informal user study provides some evidence that users prefer the natural gestural interface over conventional 3D input device interactions such as the use of input device buttons. The study also showed a strong preference for the 3D immersive city visualization over conventional 2D displays.