Analysis 
Real World 
  
  
  
Scientific/Physical Modeling 
  
Scientific/Phenomenal Model 
  
Computational Modeling 
  
  
  
Conceptual Computational Model 
  
Design 
  
Logical Computational Model 
  
  
Implemetation 
Physical Computational Model 
  
  
  
  
Figure 1 System life-cycle of GIS 
into something that really works is called the implementation 
stage. In this stage, object-oriented database management 
systems and object-oriented programming languages are 
suitable mechanisms to implement the object-oriented systems. 
When the life-cycle of GIS moves to usage and maintenance, 
software reuse and inheritance techniques can be used to 
develop and maintain user application systems. 
This paper discusses the conceptual model of geographic 
phenomena by object-oriented analysis in section 2; object- 
oriented logical spatial data model in section 3; design of 
object-oriented spatial database management engine(GeoDB) 
in section 4; software modeling implementation of an 
OOGIS software (GeoStar) in section 5; and finally a 
conclusion in section 6. 
2. OBJECT-ORIENTED CONCEPTUAL MODEL IN GIS 
Figure 1 shows several levels of the system life-cycle and 
their translation. For geographic information system, various 
phenomena on the earth’s surface are scientifically translated 
to images via photogrammetry and remote sensing, or maps 
via cartography and field surveying. The phenomena or objects 
on images or maps can be represented by an object-oriented 
model. The following semantic concepts are important in 
OOGIS. They are Object, Classification, Generalization, 
- Aggregation, Association. 
When we look at images or maps, we can identify buildings, 
parks, lakes, highways, roads, railroads, rivers, power lines, 
towers, train stations and etc., then classify them to feature 
classes. Furthermore, class building might be specialized into 
classes supermarket, post office, hotel, restaurant, house and 
so on. From the other side, we can combine classes building, 
park, lake and etc. into a common superclass surface, because 
they have the same geometric properties. Several features 
might be aggregated to a complex object, for instance, some 
buildings and other features can be grouped to a university or 
a factory. 
In geometry, there are four highest superclasses (POINT, LINE, 
SURFACE, and COMPLEX) abstracted from all features of 
GIS (see figure 2). All spatial features are defined as 
belonging to various classes. Each feature class is declared as 
belonging to a superclass of four geometry classes, which are 
distinguished according to representation but not phenomenon. 
A characteristic of such a feature class is called an attribute 
(non geometry). Certain attributes are identified with each 
class. For example, the class BUILDING has attributes like 
building number, street, zip, city, owner, usage, built date, 
floors, etc.. The specialized subclasses of a feature class might 
have additional and more detailed attributes, and can inherit 
the common attributes from their superclass. For example, 
Class HOTEL is a subclass of class BUILDING, it inherits the 
attributes of BUILDING like building number, street, zip, 
city, owner, function, built date, floor number, etc. It also has 
some special attributes such as employer , manager, room 
number, bed number, income, tax, etc. (see figure 2). 
  
Spatial Object 
  
  
  
  
  
CN, e 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Point Line Surface : Complex 
Tower Train Station Road River Building Lake Factory University o 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Figure2 Feature classification 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996 
  
3. OI 
GIS 
The « 
featur 
logica 
For Gi 
model 
data 1 
becau: 
relatio 
data o: 
A very 
betwe: 
applic 
aggreg 
Thare 
LINE, 
4. 
MAN, 
An obj 
GeoDE 
Techni 
based. 
actuall 
prograi 
the sar