虚拟群体模拟中的大规模场景简化技术

发布时间：2018-04-29 21:50

本文选题：群体动画 + 实时渲染　；参考：《电子科技大学》2010年硕士论文

【摘要】： 大规模群体动画一直是虚拟现实技术研究的热点之一。群体动画技术广泛应用于游戏、电影动漫、建筑规划以及火灾等灾难现场模拟与辅助指挥。通过群体行为的真实模拟,可以辅助电影动漫产生真实宏大场景效果;可以在公共场所(商场、体育场馆、娱乐场所等)规划设计中模拟验证安全通道等设计的合理性;可以为智能交通管理(红绿灯有效设置,主干路口的交通流量管理)提供模拟环境;也可以为灾难现场指挥提供指导和训练平台。目前这个方面的研究得到了广泛的重视,对群体动画的研究和创作工具产生了极大的需求,同时一套完整高效的群体动画设计与制作工具也将促进影视动画等方面数字媒体产业的发展。作为群体动画技术的一个关键技术,模型简化占据着不可替代的重要位置。要实现大规模的场景,必然会涉及要绘制和渲染大量的模型,而对于实时场景渲染来说,渲染速度和模型的视觉逼真度是两个非常重要的指标。要达到两个指标的和谐统一,那么就必须选择一个很好的算法来实现渲染速度和模型显示逼真度的折中。对于静态模型,现有技术已经很好的实现了模型简化。根据简化算法的简化方式可将其分为四大类:顶点聚类算法、增量式简化算法、采样算法和自适应细分算法。在这四类模型简化算法的基础上发展起来的静态大规模场景简化技术也比较完善,如基于细节层次模型的大规模场景简化技术和基于点采样的大规模场景简化技术等。本文就以上算法和技术做了具体的研究和分析,并生成了一种新的算法:一种基于点采样的模型简化算法和大规模场景简化技术。虽然静态大规模场景简化技术已经比较完善,但动态的大规模场景的简化技术才刚起步,实现效果也不理想。针对这种情况,在提出的新算法中,加入了动态模型的简化新方法。实验结果表明:在保证场景的渲染速度和视觉真实性的条件下,能很好的实现30000人的实时动画。
[Abstract]:Mass animation has always been one of the hotspots in virtual reality technology. Group animation technology is widely used in game, movie animation, architectural planning and fire disaster scene simulation and auxiliary command. Through the real simulation of group behavior, it can assist the movie animation to produce real and grand scene effect, and can simulate and verify the rationality of the safety channel design in the planning and design of public places (shopping malls, stadiums, entertainment places, etc.); It can provide a simulated environment for intelligent traffic management (traffic lights and traffic flow management at trunk junctions), as well as a guidance and training platform for disaster scene command. At present, the research on this aspect has received extensive attention, and there is a great demand for the research and creation tools of group animation. At the same time, a set of complete and efficient group animation design and production tools will also promote the development of digital media industry. As a key technology of group animation, model simplification occupies an irreplaceable position. In order to realize large-scale scene, it is necessary to draw and render a large number of models. For real-time scene rendering, the rendering speed and the visual fidelity of the model are two very important indicators. In order to achieve the harmony and unity of the two indexes, we must choose a good algorithm to achieve the compromise between rendering speed and model display fidelity. For static models, the existing techniques have achieved model simplification well. According to the simplified algorithm, it can be divided into four categories: vertex clustering algorithm, incremental simplification algorithm, sampling algorithm and adaptive subdivision algorithm. The static large-scale scene simplification technology based on these four models simplification algorithms is also relatively perfect, such as the large-scale scene simplification technology based on detail hierarchy model and the large-scale scene simplification technology based on point sampling. In this paper, the above algorithms and techniques are studied and analyzed in detail, and a new algorithm is generated: a model simplification algorithm based on point sampling and a large-scale scene simplification technique. Although the static large-scale scene simplification technology has been relatively perfect, but the dynamic large-scale scene simplification technology has just started, the realization effect is not ideal. In order to solve this problem, a new simplified method of dynamic model is added to the proposed algorithm. The experimental results show that under the condition of guaranteeing the rendering speed and visual authenticity of the scene, the real-time animation of 30000 people can be realized very well.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2010
【分类号】：TP391.41

【引证文献】