中文摘要：

在研究地形多分辨率细节层次模型构建理论的基础上,提出了结合地形特征的模型细节层次的细分过程,改进了基于外存的数据存储设计方法。设计了基于增量段式存储的数据组织方式,研究了地形数据绘制过程中的裂缝现象,引入模型高度损失误差计算方法,通过采用增删高程点的办法,消除了绘制图形中的裂缝。以某地区的DEM数据为实验数据,采用C＋＋和OpenGL编程实现了地形的绘制。结果表明：本存储策略优于传统存储策略,通过增量段式的形式存储不同分辨率的高程点,重叠的高程值不再重复存储,以节约外存空间,减少了数据冗余度,I/O操作时间提高率为25%。且随着数据量增加,I/O操作时间提高效率更为显著。该存储设计将物理分块、逻辑分层、索引检索方法结合在一起使用,在减少外存占用率、降低数据量加载的同时,也提高了数据扩展的灵活性。当增加地形分块时,不会影响原有分块和存储结构。实验效果与原始地形非常相似,保证了生成的细节层次模型的真实性。此外,针对DEM实验数据,在加载非相邻分辨率等级层次的数据时,地形绘制过程出现裂缝的问题,通过删除高程点的办法来消除裂缝,取得了很好的实验效果。

英文摘要：

This article improved data storage design based on external storage when building the details of multi-resolution hierarchy model. However, it also designed a special organization model based on incremental Segments. Meanwhile, by studying the crack phenomenon in the process of terrain rendering, the model error of height loss calculation to solve this problem of cracks has been introduced in this article. Finally, the effectiveness of the model and storage strategy has been verified in this paper with the data of DEM in a region as the experimental data. And it also achieved the terrain rendering through the programming language of C ＋＋and Open GL. In this article, the storage strategy, which is superior to the traditional storage manner, can save the storage space peripheral（SSP）. That is a manner for incremental sections storage to elevation point of different resolution and non-overlapping storage to elevation values. By this strategy, it can reduce the amount of data redundancy and improve the I/O operation time（improving rate is 25%）, and more the operating time efficiency can be significantly improved with increase of data. The storage design not only reduces the SSP occupancy rate and the amount of data loading, but also improves the flexibility of data extension. Moreover, there is no affect to the original block and storage structure when increasing the terrain block. Finally experimental effect is very similar to the original terrain, which guarantees the authenticity of the level of detail model. In addition, the terrain rendering process, considering the DEM data, will make crack when loading the non-contiguous resolution data hierarchy. Concerning this issue, the wonderful experiment results can be obtained by eliminating the cracking in the manner of removing the elevation point.