Representation of 3D City Models Based on Two Interactive Approaches in Mobile Devices

Repre­­sentation of the user's environment along with functionalities like route finding and exploring point of interests (POIs), is usually provided in the form of 2D maps which limit the user to a top-down view of the nearby environment. Considering the 3D form of real-world objects, the presentation of them as 3D city model plays an important role in visualization and location-based functionalities. In the visualization part, 3D city model helps users to identify real-world objects by making a virtual environment and regarding the geometric attributes of them. Furthermore, 3D city model is utilized in variety of applications such as visibility analysis, energy demand estimation, urban facility management, 3D cadastre and indoor navigation. Recent improvements in the hardware and software specifications of smart phones along with the utilization of powerful CPUs and GPUs provided the developers with the ability of exploiting mobile devices as platforms for different location-based usages. Virtual Reality (VR) is a technology that simulates objects of the real world in a virtual environment and enables users to interact with those objects. 3D city models that contain elements like buildings, vegetation areas, roads and terrain, are the examples of VR applications in smart phones and tablets. This paper describes a mobile-based VR application to represent and explore campus objects exists in the faculty of Geodesy and Geomatics Engineering of K.N.Toosi University of technology. Generally, OpenGL ES API was used to render 2D and 3D graphic objects in mobile devices and hardware accelerated by using GPU. In spite of that, 3D city model elements like buildings are complicated objects, which produce in desktop graphic softwares such as Google Sketchup, City Engine, 3ds MAX. A variety of game engines and graphic libraries are developed to load and render graphic models based on the low-level structure of OpenGL ES. These libraries provide functionalities like loading and creating objects, supporting different graphic formats, collision detection and so on. Our mobile application is developed based on Libgdx cross-platform game engine. 3D models of campus objects are produced in Google Sketchup software form 2D footprint of buildings in 1:2000 aerial map. In the next step, models were textured and their coordinates and non-spatial attributes were stored in a KML file format. The serialization process was performed in the mobile device to construct and render models in graphic environment. In this study, we have implemented two types of representation and interaction with 3D city model: 1) Birds-eye view and 2) First-person view. In the Bird's eye view, user is elevated above the surface and control buttons and touch screen events handle user's interactions. On the other hand, the first-person view is mimic of Augmented Reality (AR) view and user can walk in the environment and explore nearby objects. In this approach, the position and orientation of the mobile device have determined by sensory data and the low-pass filter has been utilized to decrease the noise of the data. In order to represent non-spatial information of objects, we have utilized frustum-culling concept to determine the selected model by the user and retrieved corresponding attributes. For that, we created a minimum bounding box (MMB) of each model and intersected pointing vector of the observer with MMBs to select nearest models to the user. The mixed interactive approach of our research enables users to explore their nearby urban environment in two different view and gets related attribute information of models directly.