三维城市模型数据划分及分布式存储方法
作者简介:李朝奎(1967-),男,博士,教授,研究方向为三维GIS建模及应用。E-mail: chkl_hn@163.com
收稿日期: 2014-07-15
要求修回日期: 2015-07-26
网络出版日期: 2015-12-20
基金资助
国家自然科学基金项目(41271390、41571374)
卫星测绘技术与应用国家测绘地理信息局重点实验室开放基金 项目(KLAMTA-201406)
Research on Three Dimensional City Model Data Partitioning and Distributed Storage
Received date: 2014-07-15
Request revised date: 2015-07-26
Online published: 2015-12-20
Copyright
随着信息获取技术的快速发展,地理信息数据每天以TB级的数量增加。三维城市模型数据作为三维GIS的重要内容,在数字城市和智慧城市建设过程中发挥重要作用。由于三维城市模型数据结构复杂,其数据量具有海量性,因此,高效地对三维城市模型进行划分及存储,以满足数据的长效管理及三维GIS系统的快速可视化数据调度和空间辅助决策需求,成为近年的研究热点。以往的数据划分方法导致划分区域在数据调度中变化频繁,使数据更新和管理变得困难,需寻找一种更为稳定且具有普适性的数据划分方法。本文分析了现有三维城市模型数据划分方法的不足,提出了基于拓扑关系模型的大比例尺图幅划分方法,并对划分后三维模型数据进行统一命名编码;借助非关系数据库MongoDB强大的海量数据组织及高效的多并发访问功能,构建了MongoDB分片集群服务器;对三维城市模型数据进行了单元划分,并采用规则建模软件City Engine进行建模,得到三维城市模型,借助非关系数据库软件MongoDB进行数据存储实验。结果表明,基于拓扑关系模型的大比例尺图幅划分方法适用于三维城市模型数据划分,划分后数据的存储效率明显提高,MongoDB数据库的多并发访问效率具有良好的稳定性。
李朝奎 , 严雯英 , 杨武 , 陈果 . 三维城市模型数据划分及分布式存储方法[J]. 地球信息科学学报, 2015 , 17(12) : 1442 -1449 . DOI: 10.3724/SP.J.1047.2015.01442
With the rapid development of information acquisition technology, the geographic information data is increasing at the magnitude of terabyte every day. As an important content of 3D GIS, 3D city model data plays an important role in the construction of digital city and smart city. Because the data structure of 3D city model is complex and the data volume is huge, how to efficiently divide and store large amount of 3D city model data in order to meet the long-term management of data, the rapid visualization of data scheduling and the requirement of spatial assistant decision-making of 3D GIS system, has become a research hotspot in recent years. Previous data partitioning methods have caused the changes of zoning frequently in the data scheduling, which makes the update and management of data become more difficult. So, it is necessary to find out a more stable and universal data partitioning method. In this paper, based on the research of the shortcomings for the existing 3D city model data partitioning methods, we proposed the large scale map partition method based on topology relation model, and then we designed a unified name encoding scheme for the 3D models data after splitting. With the help of the powerful massive data organization and efficient multiple concurrent access function of the non-relational database MongoDB, a MongoDB sharded cluster server is constructed. The 3D city model data was used in unit division, and the rules modeling software City Engine was applied to processing the divided units, thus producing the 3D city model. Afterwards, MongoDB was used for data storage experiments. The results show that the large scale map partition method based on topology relation model is capable and sutable for the data partition of 3D city model, and the storage efficiency of the divided data is obviously improved. Moreover, the MongoDB database has a good stability on multiple concurrent access.
Key words: 3DCM; data partition; spatial topological relations; spatial database; MongoDB
Fig. 1 The hierarchical diagram of B-REP model图1 B-REP模型分级图 |
Fig. 2 The Octree model图2 八叉树(Octree)模型 |
Fig. 3 The geometry spatial data model of 3DCM图3 3DCM几何空间数据模型 |
Fig. 4 The modeling unit division based on 2 km×2 km map sheet图4 基于2 km×2 km图幅的建模单元划分 |
Fig. 5 Four kinds of basic relationship between object and map图5 地物与图幅的4种基本关系 |
Fig. 6 The organization chart of 3D model图6 三维模型组织结构 |
Fig. 7 Rule-based modeling procedure of CityEngine图7 基于规则的CityEngine建模流程 |
Fig. 8 The construction achievements of 3DCM图8 构建的三维城市模型 |
Tab. 1 The cluster environment of MongoDB表1 MongoDB集群环境 |
服务器编号/IP | 路由服务器端口 | 配置服务器端口 | 分片/端口 |
---|---|---|---|
1:192.168.0.1 | Mongos1:10000 | Configdb1:20 000 | Shard1:27 001 Shard2:27 002 |
2:192.168.0.2 | Mongos2:10000 | Configdb2:20 000 | Shard1:27 001 Shard2:27 002 |
3:192.168.0.3 | Mongos3:10000 | Configdb3:20 000 | Shard1:27 001 Shard2:27 002 |
Tab. 2 The comparison of image data storage time表2 影像数据入库时间对比 |
数据量(MB) | 时间(s) | |
---|---|---|
Mongo DB | SQL Server 2005 | |
500 | 98.231 | 173.962 |
2000 | 327.917 | 631.674 |
4217 | 745.483 | 1752.790 |
Tab. 3 The data storage time comparison of different methods表3 不同处理方法数据入库时间对比 |
模型种类 | 数据大小(GB) | 入库时间(s) |
---|---|---|
① | 11.7 | 1257.695 |
② | 12.1 | 1309.553 |
③ | 20.5 | 2478.391 |
Fig. 9 The time efficiency comparison chart of simulating multiple concurrent access图9 模拟多并发访问时间效率对比 |
Fig. 10 The comparison of efficiencies for multiple concurrent access with different processing methods图10 不同处理方法的数据多并发访问效率对比 |
The authors have declared that no competing interests exist.
[1] |
|
[2] |
|
[3] |
|
[4] |
|
[5] |
|
[6] |
|
[7] |
|
[8] |
|
[9] |
|
[10] |
|
[11] |
|
[12] |
|
[13] |
|
[14] |
|
[15] |
|
[16] |
|
/
〈 | 〉 |