飞行监控系统中基于GPU的地形渲染

利用实时遥测飞行信息生成三维虚拟地形场景,旨在飞机试飞时,增强地面指挥人员对飞机实时位置地面场景的感知能力,帮助飞行员在试飞时处理复杂多变的飞行状况。为了逼真且流畅地生成三维地形场景,首先使用固定网格映射(PGM)的方法生成地形采样网格,接着利用几何着色器为三角形面片分配连续的与视点相关的细节等级(LOD),对缓存在纹理数组中的各等级高程数据和卫星照片进行混合采样。由于细节等级是连续的,因此各级纹理和高程之间能够平滑的过渡,增加了场景的真实性。最后,考虑到显存的容量限制,引入了一种内存与显存之间的纹理调度方法。整个算法都是面向图形处理单元(GPU)实现的,能够较好地兼顾实时性和真实性,实现多分辨率大规模地形场景的实时虚拟显示。根据本文算法实现的实时监控系统在支线飞机的飞行试验中取得了较好的效果。

GPU-based Terrain Rendering in Flight Monitoring Systems

The purpose of this article is to enhance the sensing capability of ground navigation officers through three-dimensional virtual terrains generated by real time remote information of aircraft status, so as to enable them to guide pilots in dealing with complex flight conditions appropriately. To create a virtual terrain, firstly a coarse sample mesh is generated by persistent grid mapping (PGM) in the algorithm. Then a continuous level-of-detail (LOD) factor is assigned to each triangle respectively based on its distance to the view point in the geometry shader. Afterwards, the factor is used to choose and blend proper levels of satellite images stored in a two-dimensional texture array in the pixel shader. Because of its continuity, the transition between different levels of textures can be smooth. Thus, the scene looks more natural and realistic. Finally, considering the capability limitation of the memory on a video card, a dynamic loading strategy is employed to manage the textures. Because the simulation is geared toward high utilization of the graphic processing unit (GPU), large-scale terrains of multiresolutions can be generated interactively according to the demands of both real-time and reality. The algorithm in the presented form is used operationally within a flight monitoring system to support the flight test of the Advanced Regional Jet at flight laboratory.