Automatic path generation for terrain navigation

摘要

Navigation in 3D terrain is considered to be a challenging task and requires virtual camera control skills such as zooming, panning and tilting. Novice users can easily get distracted and disorientated which may result in being lost in space. Solutions for the virtual environment exploration problems are still being researched in order to assist users during their journey inside virtual environments. Among these solutions, assisted camera control techniques require viewpoint computation and path planning. This paper introduces a novel approach to navigate over a 3D terrain using best viewpoints. We exploit the concept of Viewpoint Entropy for best view determination and use Greedy N-Best View Selection for visibility calculations. We integrate road network data to extract regions for detailed visibility analysis in subsections of the terrain. In order to connect the calculated viewpoints, we use an evolutionary programming approach for the Traveling Salesman problem. We present the generated tour using a Google Earth framework. The computed and planned viewpoints in our solution reduce human effort when used as starting points for scene exploration or to generate the representative images of the terrain dataset. The proposed framework can be integrated into 3D game engines or urban visualization systems. This integration gives quickly a glimpse or tour of the environment for novice users without the help of prior planning. Furthermore, terrain visibility used together with road network data and an optimization method for final path construction allows the computed path to be used for large urban area reconnaissance and surveillance tasks with aerial vehicles.