nbul 2004 
ard. It is 
etermined 
map. The 
. of road 
bute data. 
ances and 
f the road 
cially in a 
Itersection 
iction, but 
ake Hong 
r network 
1 as many 
mount of 
lidation in 
associated 
nt well the 
at contain 
and Shaw 
over-pass 
hich does 
trict those 
impedance 
inefficient 
always the 
hey could 
^ time and 
>, the most 
S Is to use 
es directly, 
lata set for 
that digital 
definition. 
ents, these 
to support 
rovides an 
TRIANS 
atively be 
al features 
/ gcometry 
without the 
r or digital 
se map and 
d database. 
national or 
However, 
the type(s) 
These will 
sed next. 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
   
     
1:5000) 
Features Descriptions 
  
  
  
  
  
  
  
  
ID Unique Identifier 
Street Blocks x Building Blocs. V 
Type barriers or traffic 
islands etc. 
Building Blocks ID Unique Identifier N 
ID Unique Identifier 
Links Lene The Length 
(Bridges, zebra- Poly | ag S Ly aon N 
crossings etc.) - 
The second polygon 
Poly2 HE 
= ID 
ID Unique Identifier 
Roads Type Footbridges, steps, N 
7 subways etc. 
  
  
  
  
  
  
Figure 1(b): Database descriptions 
It is clear that the movement of cars is limited to one particular 
linear feature road only. But for pedestrians, they are 
supposed to walk freely on all walking features (like pavements, 
stairs, clevated walkways) and on restricted places of roads (eig. 
zebra crossing, traffic light areas). To start with, street block 
layer is chosen to be the key layer that presents the basic 
movement of pedestrians. Since other walking features such as 
steps, buildings, and pavements are all inside street blocks, the 
movement within street block is not our main concern here. The 
second important features to be considered are the “links” 
between street blocks. These “links” could be zebra-crossings, 
bridges or subways that enable pedestrians to move from one 
street block to another 
  
  
  
  
  
  
  
  
  
Figure 2: Example of computing a walking path between À 
and B 
Suppose a walking path has to be computed between source A 
and destination B (Figure 2), the proposed algorithm is to 
construct a theoretical shortest line connecting these two places. 
All calculations and and analyses are performed with reference 
to this theoretical line. This line is a vector which contains not 
only magnitude but also direction. The source is always 
considered as a from-node whereas the destination is considered 
as a to-node. After the direction is determined, the line is then 
broken down into several segments by extracting intersection 
points between the street block and “link” features. In the same 
way, each segment is considered as a directional line where the 
direction is the same as the drawn line. By overlaying the start 
and end point of each line segment with the polygonal strect 
block layer, a list of polygons ID is resulted for further 
calculations (Figure 3). 
  
  
  
  
  
  
  
  
  
  
Segment Polygon 1D 
From | To 
| = Polygon 
1 3 3 3 y? 
2 3 1 3 
3 ; 3 "s 
A 
4 ui UM 2 
8 2 2 2 
  
  
  
  
  
Figure 3: Polygon List 
The resulting list indicates that to walk from A to B, the 
pedestrian is suggested to start at polygon 3, polygon 1 and then 
polygon 2. Then the next question is: how to walk across 
polygon 3 and 1, then polygon I and 2 on ground? To cross the 
road we need to find some features that connect the two street 
blocks, such as subways, bridges or zebra-crossings. Hence, the 
next step is to obtain the connectivity spatial relationship 
between the ‘link’ and street block features. This can be done 
by querying the attribute table of the “link” (Figure 4). 
B3 Attributes of links 
    
  
  
  
  
  
   
Figure 4: Attribute table of links