可扩展的实时自然景物模拟算法

针对传统的粒子系统实时仿真存在只能针对单一自然景物模拟、计算耗时、图像不真实、算法复杂等问题，提出了一种基于粒子系统和图形处理器（GPU）加速通用可扩展的自然景物模拟算法。在该算法中，粒子的物理运动计算过程和渲染阶段完全由CPU转移至GPU，可以增加粒子数量和提高渲染速度；同时，在渲染过程中，可以较好地利用硬件支持的粒子图技术来改善渲染中粒子的外表，选择不同纹理，从而能够较方便地模拟不同的自然景物。最后，在GPU上实现了雪花、喷泉、烟花、瀑布等模拟，算法充分利用了GPU的多通道并行处理性和可编程性，提高了自然景物模拟的实时性，可运用于虚拟现实系统。

A scalable real-time natural scenery simulation algorithm

Conventional particle simulation methods have some limitations, such as singleness, time consuming calculation, low reality, algorithm complexity, etc. In view of this problem, a novel scalable real time natural scenery simulation algorithm is proposed, which system, which is based on particle system and Graphics Processing Units (GPU). In the proposed algorithm, the computation process and rendering of particles’ physical motion is on GPU rather than CPU, thus increasing the number of particles and improving the rendering speed largely. Meanwhile, in rendering process, the appearances of particles are improved by the particle graph technique with hardware support, and different textures are selected to facilitate simulating different sceneries (rain and snow, fountain, firework, waterfall). Finally, these sceneries simulation on GPU shows that our algorithm makes use of the parallelism and programmability of GPU, improves the real time requirement of natural scenery simulation and can be applied in virtual reality systems.