International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 
  
2.4 Modeling the physical environment 
Researchers have been known to simulate the real world using 
virtual reality or augmented reality in a computerized setting 
(Huber and Sieber 2001, Nagao and Katsuno 1998). The 
backdrops for some applications have been elevated from two- 
dimensional (2D) to three-dimensional (3D) or even multi- 
dimensional (plus time, for instance) setting (Elvins et al 1999, 
van Dijk et al 2003). 
Shiode (2001) did a thorough review on the current 
development of 3D digital urban models. Three categories of 
models were summarized in his article. The first approach 
utilized various methods for capturing heights and facade 
information at the data input stage. The second approach 
combined utility and analytical features from different 
platforms to yield the desired functionality. The third group was 
derived from the amount of geometric content for the degree of 
reality. 
However, not all study favors the multimodal over the 2D 
approach.- Draper (2000) showed in an empirical study that a 
2D map is easier to understand than its 3D counterpart in a 
wayfinding cognitive test. A 2D approach augmented with 
multimedia addition might be a reasonable alternative in view 
of cost and time constraints. 
2.5 Presenting route guidance 
The presentation of route directions and instructions hinges on 
a few conditions: 
* Time of usage 
L] Advanced planning and search for route directions 
[J Accompanying or enroute directions 
= Medium of delivery 
L] Speech or verbal instructions 
Ll Text or written instructions 
= Format of instruction 
[] Street-based instructions 
L] Landmark-based instructions 
L] Cardinal direction-based instructions 
L] Time-based instructions 
In considering time of usage, accompanying or enroute 
directions are possible only if the user's present location and 
orientation can be determined. With regard to the medium of 
delivery, text or written instructions are unambiguous and 
understandable by the general public while speech or verbal 
instructions can supplement at a later stage. We would prefer 
andmark-based instructions as the format of instructions 
;»ecause many footpaths in our large-scale setting are missing 
names and the remaining options not favored in Hong Kong. 
3. Case Study 
3.4 Background 
Situated at the mid level of a mountainside, the University of 
Hong Kong (HKU) has a main campus congested with high- 
rise buildings. Campus directories in two-dimensional 
representation (Figure 1) are posted close to the gate entrances 
where the majority of visitors start their journeys. They are 
L9 
{D 
also erected in open spaces where people tend to gather or stop 
for a rest. This directory system is confusing for a complex 
campus packed with buildings not only distributed on hilly 
terrains but also connected with corridors or steps at different 
elevations. 
   
  
  
  
  
MET ATE 
Fm AR Ve 
ENTRANCE 
  
    
Figure 1: Main Campus of the University of Hong Kong 
3.2 Data construct for network analysis 
The spatial environment of the HKU campus is characterized as 
follows: 
= Does not necessarily possess unique reference points in 
space 
A visitor may start from any place on campus, walk along 
a path, through a building, and continue to other 
destinations situated on campus. Hence the routing 
exercise may not be limited to only one modeled setting. 
" Naming system is missing for certain routes 
All public roads and streets hold official names. Private 
owned roads are sometimes unnamed. This naming gap 
will cause difficulties in street-based route instructions. 
" Walking paths for necessarily 
represented 
pedestrians are not 
The campus road network can be derived by extracting 
centerlines from existing roads in a digital map. However, 
footpaths for pedestrians are often overlooked. These 
unofficial paths are often preferred over regular roads in 
campus navigation and a comprehensive field survey may 
be needed to verify and record such footpaths. 
3.3 The modeling procedure 
Five components have been identified as essential to the data 
construct. 
Node 
* Building Entrance/Exit points: 
These include entrances located on the ground level of 
cach building and alternative entrances into a building 
which may be located on other floors of the building. 
Interr 
Link 
A lir 
pedes 
the s 
Most 
Wher 
Walk 
move 
becau 
Land 
Landi 
(196€ 
acces 
proto 
The 1 
geom 
Psych 
of a1 
one o 
availa 
comp 
Data 
The 
Cente 
data € 
buildi 
consti