随着地球空间数据获取技术的进步与三维地理信息系统的发展，三维地理环境模型的几何与纹理数据量呈几何级数增长，然而大众化PC机有限的CPU/GPU处理能力与存储空间、受限的传输速度与带宽，导致大规模三维模型数据的实时可视化变得十分困难。本申请立足于单机环境大规模三维模型数据的实时可视化需求，研究数据组织管理与绘制优化的数据粒度方法，包括：基于数据内容与多源特征的粒度表达、多尺度空间粒度表达、多层次存储与调度粒度表达、实时可视化任务处理代价粒度表达，建立数据粒度模型并研究粒度的量化标准与粒度间的等价转化关系；研究基于数据粒度模型的组织与管理方法和实时可视化流水线中数据处理代价全局优化方法；以真实三维城市模型数据作为基础实验数据，验证数据粒度模型的正确性与方法的有效性。该研究通过粒度将多源数据内容分类、空间组织、存储硬件和可视化处理任务等有机联系起来，为三维空间数据的实时可视化提供了新的有效方法。

With the great improvement of data acquisition technology and the development of 3D city modeling, the data volume of the geometry and texture grows geometrically and exponentially. The real-time rendering applications of large-scale 3D city models face great challenges, because of the limited memory and narrow bandwidth of computers. Aiming at real-time rendering of massive 3D models, this project proposal a new data organization schdule and rendering optimization method based on data granularity model, which including the following aspects: the represent of data granular based on data content and multi-resource acquisition, the represent of data granular of multi-scale spatial organization, the represent of data granular of multi-levels data storage and schedule strategy, the represent of data granular of rendering cost of the tasks in visualization pipeline. The data granularity model has been proposaled based on these four facts, and then the quantity method of granular and equivalent relationship between the granulars have been studied.A new data organization and management method and a global optimization data schedule method have been developed. At last,this proposal using real models of Wuhan city to test and verify the data granularity model.The project make a combination of the data content, spatial organization, computer hardwards and rendering tasks through granularity, which contributes a new method for large-scale 3D data real-time applications.