近界视景导航中森林场景加速绘制

作为近界视景导航中典型地物,三维几何树模型能提高虚拟场景视觉精度,增强场景真实感,但会导致大规模森林场景绘制效率降低,从而降低视景导航精度。为提高绘制效率以确保视景导航精度,并提高场景真实感,采用布告板云算法构建了基于视距的多分辨率树模型,针对算法在近界视景导航应用中存在的问题进行了改进。为进一步提高近界特别是贴地飞行时高动态视点下视景导航精度,根据有限视域、威胁区域以及视觉冗余等特点,提出了一种基于高动态视点的加速绘制算法。通过实验对比分析,算法大幅提高了场景绘制效率,增强了场景视觉质量,提高了虚拟场景导航精度。

Accelerating rendering of the forest scene in the near-ground visual navigation

As the typical ground objects in near-ground visual navigation application, the three-dimensional geometrical models of trees can improve the visual precision and the reality of visual environment. However, it could lead to the decrease in rendering efficiency of a wide forest scene and results in loss of navigation precision. In order to boost the rendering efficiency for ensuring the navigation precision and maintain the sense of reality, billboard clouds algorithm was employed to establish the multi-resolution representations of a tree model based on view distance. To improve the navigation precision further with high dynamic viewpoint in the near-ground especially the nap-of-the-earth flight, according to limited field of view, threatened area and visual redundancy, an accelerating rendering algorithm based on high dynamic viewpoint was proposed to improve the rendering efficiency while high-fidelity scene was kept. Through contrast analysis of experiments, the proposed algorithm has high-efficiency rendering and high-fidelity scene, which is appropriate for the high-precision demand of the near-ground visual navigation.