anbul 2004 
al., 1998) 
> 
i 
Squares 
ustment 
A 
rpolate Position 
7 
    
ise IMU Data 
¢ 
bsystem 
process of 
s (or other 
to the data 
yement of 
inition of 
activity of 
ainly with 
| mapping 
view of a 
del and a 
es’ of the 
les, streets 
| buildings 
traditional 
on paper 
rocess as 
> VIEWS. 
subsystem 
shows text 
owing the 
ing sensor 
  
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV , Part B2. Istanbul 2004 
Graphical User Interface (GUI) is another aspect to be 
improved in the automated mapping system using software 
engineering technology. An improved GUI version of the 
bundle adjustment could be as shown in Figure 4, where the 
control points, image boundaries are graphically demonstrated, 
instead of showing them as texts and numbers. 
   
Pa YA Bn 20ER *ZBLQU AGBERE GAR 
z HESS ;* T 
EASE p* m 
detur tetti veiut emiten rete roti tt pete cepe AT 
OCICS 
Ri WI i T 
pongpiaoarnm 
  
  
— 
8 % i 
E me — 
I, 
  
  
  
Figure 4. Improved GUI for bundle adjustment 
3. UML BASED DIGITAL TERRAIN MODELLING 
SYSTEM 
In this section we describe the development of the system that 
utilizes UML as a modeling tool. The digital terrain system is 
based on DTM data structure called Triangular Irregular 
Network (TIN). It is a popular structure in DTM domain — 
offers several advantages over other structures like grid. One 
of the advantages is it can accommodate linear features that 
exist on the terrain quite well. ^ Abdul-Rahman (2000) 
constructed a system based on the Booch modeling tool 
(Booch, 1994) and could be extended by using the UML as 
describe below. 
In this section we discuss the use of UML for constructing 2D 
and 3D TIN. First, the 2D TIN — the module has three major 
operations, they are Distance Transformation (DT), Thiessen 
polygonization, and 2D TIN topology. Figure 5 shows the 
class diagram of the operations. 
  
Point2Raster 
  
   
  
   
2D- 
3DTINStrcturing 
  
  
  
Arrasterize(} 
  
   
  
/ 
  
  
2DTINConstruct 3DTINConstruct 
  
  
  
  
I-3DDistanceTransform() 
+30 ThiessenPolygon() 
+ 3DTINTopology() 
+DistanceTransform() 
+ThiessenPolygon(} 
+2DTINTopology() 
  
  
  
  
  
  
Figure 5. The UML class diagram for the 2D and 3D TIN data 
structuring. 
The results from the 2DTINConstruct and the 
3DTINConstruct classes, i.e. the 2D and 3D TIN structured 
data become an input to the Digital Terrain system. The 
modelling and the development of the system is discussed in 
the following sections. Here, we provide a discussion of the 
OO TINs spatial data modelling techniques. Conceptually, the 
200 
general modelling steps (the conceptual, the logical, and the 
physical) could be used for TIN spatial data modelling. 
Others called it as the three-step approach. The class schema 
for spatial data modelling are described below. 
3.1 The Class Schema 
The schema is based on several classes, they are Spatial 
Objects (the super class), and four major subclasses which are 
Node, Edge, Polygon, and Solid. 
Spatial objects 
The spatial object class is a general class of the real world 
objects. It is the super class in the class hierarchy. We 
assume that all other objects are derived from this super class. 
All terrain objects could be categorised into several sub 
classes such as points, lines, areas, and solids (volume) 
features. In OO modelling, these feature types are the classes 
in the modelling hierarchy. 
Node 
A node can be considered as the most basic geometrical unit in 
spatial data modelling. It may represent point entities or point 
objects at a particular mapping scale. Examples of point 
objects are wells, terrain spot heights, and the like. 
In geoinformation, we may represent these objects by a class 
called a node class. The coordinates of the nodes (including 
the nodes represent edges) are held by a coordinates container 
class, called XYZContainer class. 
E 
OOTINSpatial 
Modelling 
  
  
UserGUI 
  
  
showt) 
  
  
  
  
  
aui) 
  
  
NodeAttribute 
——J 
  
  
  
Geometric 
T A+point() ArcContainer 
2 büne( 
| put Esurfacen m. GelArCIDO 
! Pad *solid(; = - 
A 
  
  
  
  
  
  
  
      
    
XYZContainer 
pr 
fav 
| " | 
" i ti 
Lem mui 
Go Ee] | 
Node i i 
| | 
i | 
| 
| 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
eftPoly() 
  
  
| 
Figure 6. UML class diagram of the system functions. 
Edge 
An edge can be represented by two nodes at each end (i.e., a 
start node and end node). In this study we consider two end 
points make a straight edge. We used this edge type to 
represent linear features. The arc container class, called 
ARCContainer holds all the arcs. The arcs container also 
serve any other class which requires arcs data in their 
operations for example the polygon class needs the arcs in 
order to form polygons. 
Polygon 
A polygon (sometimes known as a surface) is used to 
represent area features such as lakes, ponds, etc. A polygon 
may be constructed by chains of closed edges. 
Solid (or Body) 
This is a representation for solid or body features such as 
buildings, trees. A chain of points and lines form body objects 
for example, 3D TIN can be represented by a series of triangle 
nodes and edges. 
The class schema in Figure 6, depicted using UML notation is 
the representation of the TIN spatial data model. The schema