kazuhiko hamamoto dept. of information media technology,
DESCRIPTION
Spatiotemporal Information Processing No.6 Data expression and Application to GIS (Geographic Information System). Kazuhiko HAMAMOTO Dept. of Information Media Technology, School of Information and Telecommunication Eng., Tokai University, Japan. Today’s Contents. - PowerPoint PPT PresentationTRANSCRIPT
Spatiotemporal Information ProcessingNo.6Data expression and Application to GIS (Geographic Information System)
Kazuhiko HAMAMOTO
Dept. of Information Media Technology,
School of Information and Telecommunication Eng.,
Tokai University, Japan
Today’s Contents
Point of view for spatial information Structure of spatial information Basic techniques of spatial information
processing Level classification Spatial inference Spatial index Spatial query
Point of view for spatial information - 1
Its expression is not “algorithmism” but “dataism”.
Spatial information includes not only “visible information” but “semantic information”. “There is a bread shop along right side of the
road which is 50m far from next cross road.” How to express “semantic information” in
spatial information ?
Point of view for spatial information - 2 How to express “semantic information” in
spatial information ?
Spatial information is expressed as “spatial structure of knowledge of the real world” and input it to a computer Conventional image data is a non-structured
data, so the role or property is different. The same expression cannot be used.
Structure of spatial data
Class or Category Object
Identifier Attribute
Geometric attribute Non-geometric attribute
Relationship Spatial relationship Temporal relationship
Object ・・・
空間データの構築 クラス (class)またはカテゴリ (category):分類名 オブジェクト (object):実体
識別子 (identifier)属性 (attribute)
幾何属性 (geometric attribute) 非幾何属性 (non-geometric attribute)
関連 (relationship) 空間的関連 (spatial relationship) 時間的関連 (temporal relationship)
オブジェクト識別子・・・
Attribute
Data set which expresses an object Geometric attribute
Abstract expression of an object by simple geometrical figure (line, dot, etc.)
Non-geometric attribute Name, area, texture, etc. of an object
“Attribute” does not comprehend all of information.It is a part of information which is extracted under a condition.
Relationship
Relationship among objects Spatial relationship
For example, “An object is next to an object.”
It is expressed by “vector data”. Temporal relationship
For example, “this object was made earlier than that one”.
Concept model of spatial data
Set of objects in “mountain” class
Vector data - 1
How to express “relationship” Geometrical figure data constructed by
“dot” data dot
coordinates line
a set of dot. start point and end point are different
Vector data - 2
Geometrical figure data constructed by “dot” data region
a closed line, whose start point and end point are the same.
surface data about 2D region or the third data for (x,y) coordinates
solid 3D geometrical data 2 data or more data for (x,y) coordinates
Example of vector data
Another vector data : Topology
connection, connotation, etc. “A road is connected to a road.” “There is a region on the right side of a road.” “There is a road along the region.” “There is a building in the region.”
It is easy to restructure and recognize spatial information
Intelligent spatial information processing The shortest path problem, etc.
Topology data
Raster data
It has a value on a grid at regular intervals.
For example, image (texture), land heights data, etc. But “contour” ,which expresses the same
height is vector data
Basic techniques of spatial information processing Level classification
Information is classified to objects. This process enables to :
Spatial inference The spatial relationship is obtained by
topology Spatial index Spatial query
Rapid retrieval from huge amount of data
Spatial inference
To obtain spatial relationship “be next to”, “be connected to” and “can go
there in 5minutes”, etc. Topology data is needed
Spatial relationship cannot be expressed by only position data
Basic components of topology data Node (vertex) Link (edge)
Spatial inference
An example of topology data
Nodeshibuya= Node1003, Nodeharajyku= Node1004,Nodeyoyogi= Node1005, Nodeshinjuku= Node1006
Linkshibuya→harajuku= Link505=( Node1003, Node1004)Linkharajuku→yoyogi= Link506=( Node1004, Node100
5)Linkyoyogi→shinjuku= Link507=( Node1005, Node100
6)
Networkyamanote_line= Network303
={ Link505, Link506, Link507}
Spatial inference Addition of “region”
by network and arrival time
Inference of “Shinjuku sta. is the third sta. beyond Shibuya sta.” or “The shortest path is ・・・” is possible.
Spatial index, Spatial query
Indexing spatial data by mesh The mesh has hierarchical framework.
The standard mesh in Japan 1st mesh : 80km, 2nd mesh : 10km, 3rd mesh : 1km
Retrieve a target included in directed area Obtain sets of mesh which covers the directed area Judge whether a target in the sets of mesh is
included in directed area or not Retrieval time depends on size of the mesh
R0 : set of mesh=1, R1 : set of mesh=2R2 : set of mesh=4, R3 : set of mesh=9
R4 : although set of mesh = 4, regions in HDD are separated.
Spatial index, Spatial query how to make mesh - 1
Spatial query by “quad-tree” The area is divided into 4 parts. The division process is repeated. Finally, each part has the same number of spatial data,
which is less than a constant. Simple algorithm because division interval is constant. If distribution of spatial data varies widely, the depth of
retrieval is different by dense or sparse. Retrieval is late in a dense region
Spatial index, Spatial query how to make mesh - 2
An example of quad-tree division
Spatial query by “K-d tree” The boundary of division is changed
according to amount of data The depth of retrieval is constant. That
means retrieval speed is constant If the data is not renewed, it is effective
Spatial index, Spatial query how to make mesh - 3
An example of K-d tree division
Spatial query by “R-tree” Spatial data is connoted by the smallest rectangular
area. The rectangular area is a representative of the spatial data.
Adjacent set of the representative area is structured. If data is renewed, the depth of tree can be kept
regularly.
Spatial index, Spatial query how to make mesh - 4
An example of R-tree division and its hierarchical structure