| Currently, the vast majority of electronic map that are used in embedded devices aretwo-dimensional. Two dimensional electronic map can’t express3D(3dimensional) worldiconically, vividly and intuitively. And it’s difficult to express complex three-dimensionalgeographic information for two dimensional electronic map, such as multilayer overpass. Withthe development of the computer software and hardware technology, three-dimensionalelectronic map came into be and have been used on some powerful PC. As the limition ofslowness of the processing speed, shortage of storage space relatively and other hardwareconditions, the three-dimensional electronic map in the embedded devices is still relatively rare,but due to the good mobility of embedded devices, it will develop rapidly. Three-dimensionaldigital map data contains three-dimensional scene data used to display and three-dimensionalpath planning data used for path optimization mainly. Because the natural landscape is verymuch, artificial road network and other buildings increase rapidly year by year. It is notpossible to load all data into memory once for embedded devices, so the data ofthree-dimensional digital map is "massive" for the small capacity embedded system memory.The out-of-core data organizational and retrieval method is important factor that affectperformance of embedded system. The traditional electronic navigation system usually loaddata which may be needed into memory once and do related operations, but for the massivethree-dimensional electronic map data, these past organization and retrieval methods ofcartographic data are basically not practical. It’s difficult to realize the quick path optimizationand showing3D scene smooth in embedded devices at the same time for the traditionalmethods. So out-of-core organization and retrieval methods of3D scene data and the path planning data in embedded systems become the focus of research in this paper.This paper study the out-of-core organizational and retrieval methods of thethree-dimensional digital map data from two aspects of the3D scene data and path planningdata, mading the following work mainly:1. For the massive3D scene data, this paper divided map into equal and inerratic smallscene firstly, then divide each small scene using the improved method of octree spacesubdivision.Then encode nodes of scene octrees using the three dimensional Hilbert curve andoptimize the linear code. The node and data in each node are stored in out-of-core memoryaccording to the coding sequence. Experimental results show out-of-core retrieval efficiency,the displaying of three-dimensional scene and roaming speed impoved due to embeddedsystems selectively loading desired scene data, no longer needed to build the memorymanagement index structure of scene dynamically and the high performance of clustering ofnodes and data in them storage.2. This paper proposes a improved octree to make up the traditional octree to ignore theintegrity of entities and object-oriented octree difficult to construct scene of the much trianglesurface and few entity objects.3. For massive3D path planning data, based on organization methods of layered withpartition, for the problems of data redundancy, not according with habits of people choosing thepath and slowness of calculating path weights and so on, combining with features ofadministrative region management, the quality of the roads, transportation and geographicaldistribution and other factors, this paper proposes the organization method of partitioning inaccordance with the administrative area firstly and the polygons of road network then. Finally,established Hilbert R-tree index with clustering for block network using the improved k-meansclustering method, improving the out-of-core retrieval efficiency, reduce the times ofout-of-core I/O and time of path optimization.4. Finally, in Android, verified the effectiveness of the proposed out-of-core organizationaland retrieval method of three-dimensional digital map data in the embedded device by SQLite. |
PDF Full Text Request