This paper describes an object-oriented approach to model a set of 2D vector data, depicting points, lines, and polygons. These primitive spatial objects are embedded in the Euclidean space with the dynamic Voronoi diagram. A spatial object condensation technique is used to partition the Voronoi diagram into smaller subspaces. Each subspace defines a bounded complex object and constitutes a node of the VMO-Tree (Voronoi based Map Object Tree), a novel data model describing relationships of condensed objects. The VMO-Tree can be constructed by either top-down partition, bottom-up aggregation, or combination of the two. Features of the VMO-Tree include: representing spatial objects in different levels of detail; node containing topologically structured complex objects, some of them child nodes; being able to search vertically between nodes of the tree and to navigate horizontally in each node in a seamless fashion; loading and down-loading spatial objects contained in a node dynamically; complex objects in the tree are independent map sheets and can be operated on separately; each map sheet corresponds to a disk page and has a minimized memory occupation.
[1]
Michael Ian Shamos,et al.
Computational geometry: an introduction
,
1985
.
[2]
Peter van Oosterom,et al.
Reactive Data Structures for Geographic Information Systems
,
1993
.
[3]
M. Molenaar,et al.
Single valued vector maps - A concept in Geographic Information Systems.
,
1989
.
[4]
Michael F. Worboys,et al.
Object-Oriented Approaches to Geo-Referenced Information
,
1994,
Int. J. Geogr. Inf. Sci..
[5]
Michael F. Worboys,et al.
A generic model for planar geographical objects
,
1992,
Int. J. Geogr. Inf. Sci..
[6]
Thomas R,et al.
Voronoi Diagrams of Line Segments Made Easy * ( Extended
,
1999
.
[7]
Christopher M. Gold,et al.
Spatial Data Structures — The Extension from One to Two Dimensions
,
1990
.