CASIA OpenIR  > 毕业生  > 博士学位论文
虚拟现实中场景绘制关键技术研究
其他题名Research on Key Technologies of Scene Rendering in Virtual Reality
李俊
学位类型工学博士
导师杨一平 ; 台宪青 ; 蒋永实
2008-05-30
学位授予单位中国科学院研究生院
学位授予地点中国科学院自动化研究所
学位专业计算机应用技术
关键词形状因子 渐进网格 分布式简化 网格压缩 遮挡度 空管仿真 Shape Factor Progressive Mesh Distributed Simplification Mesh Compression Degree Of Occlusion Culling Air Traffic Controller
摘要虚拟场景的快速绘制技术是虚拟现实中的关键技术。为了真实的表达虚拟场景,人们通常采用多边形网格来构建虚拟场景。随着3D 建模技术的不断发展,三维模型精度不断提高,多边形数量也随之飞速增长,它给计算机的实时绘制带来了巨大的困难。本文主要从场景简化和快速绘制的角度来展开研究。论文的工作主要有以下几个部分: 为了保持简化模型的主要轮廓细节特征,我们提出了一种基于形状因子的方法来改进基于渐进网格的简化方法。给出了关于形状因子的数学定义,它能够很好的描述三维模型顶点的外形特征,从而克服了二次误差度量方法在低分辨率情况下丢失模型特征的缺点。我们用Metro误差检测和图像质量比较来衡量简化模型效果,实验结果表明我们的方法取得了较好的效果。 针对海量网格模型的层次细节模型自动生成,我们提出了一种基于分布式处理框架的网格简化方法。为了减少网络传输时间,我们提出了一种快速无损的三维网格压缩方法来减少数据在网络中的传输。实验结果显示我们方法提供了一种有效的快速简化框架。 从场景实时图形生成的角度出发,提出了一种结合概率遮挡的视点相关网格模型绘制方法。视点相关的简化能够实时动态的改变虚拟场景的细节层次,然而在场景中存在大量遮挡物的情况下,这些遮挡物体将有可能较高分辨率级被绘制,降低图形绘制帧率。为了克服这一缺点,我们通过快速计算区域相对视点的遮挡度,使具有较高遮挡度的区域在较低细节层次绘制,并根据帧与帧之间的相关性进行了遮挡度计算的优化,进一步提高了绘制帧率。实验结果表明了该方法的有效性。 基于场景加速绘制技术原理,将本文所介绍的关键技术应用于空管仿真系统中,从而提高系统的仿真能力。提出了建立空管仿真系统的基本框架,实现了基于分布式网络的空中管制仿真的主要功能。同时结合工程具体情况,从功能模块、结构等方面介绍了空管系统的开发流程。系统运行结果显示应用相关技术后大大提高了场景的绘制帧率。
其他摘要The fast rendering of virtual scene is a key technology of virtual reality. To express the virtual scene in natural, people use 3D model to construct the virtual scene. While the accuracy of these 3D models grows with the development of the modeling techniques, the volume of the data sets grows even rapidly. It brings great pressure to the rendering of the computer. Our work mainly focuses on the simplification of virtual scene and fast rendering. The work mainly consists of the following aspects: To preserve the major feature of the modeles, we developed an algorithm based on shape feature to improve progressive mesh simplification method. We give the definition about shape feature in detail, and it can describe the major feature perfectly, which overcome the drawback of quadric error metric method that is easy to lose the model feature under low-resolution circumstance. We employ Metro detection and image comparison to verify the error measurements. Experimental results demonstrate the effectiveness of the proposed method. As for automatic gerenating Lod model of the large-volume 3D model, we designed a distributed framework for the simplification of massive mesh. To reduce the time of processing, we developed a fast and simple scheme for lossless, on line compression method to reduce the time of transmit between the network. The validity of framework is demonstrated by the experiment results. From the point view of real-time graphics rendering, we proposed a view-dependent simplification algorithm based on probability occlusion culling. View-dependent rendering provides the ability to change level-of-detail smoothly in real-time. However, many occluded regions will be rendered in high level-of-detail, which reduced the frame rate.To overcome this drawback, we estimate the occlusion factor and fastly determine the degree of occlusion of region with respect to the viewpoint. The region with high degree of occlusion will be rendered in low level-of-detail. At the same time we take advantage of coherence between consecutive frames to improve the fame rate. Experiment result demonstrates the effectiveness of the proposed method. Based on the theory of fast rendering, we applied the key technology to the airport traffic controller simulation and validation system, the simulation ability of the system can be improved. The main framework for constructing airport traffic controller distributed simulation system is brought forward. Combined with the environment of project, the design of the system and developing flow are discussed in detail from several aspects, such as function module, configuration, etc. The result of the system operation shows that the frame rate of the scene has been improved greatly by the application of our key technology.
馆藏号XWLW1280
其他标识符200418014628012
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/6102
专题毕业生_博士学位论文
推荐引用方式
GB/T 7714
李俊. 虚拟现实中场景绘制关键技术研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2008.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
CASIA_20041801462801(4495KB) 暂不开放CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[李俊]的文章
百度学术
百度学术中相似的文章
[李俊]的文章
必应学术
必应学术中相似的文章
[李俊]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。