Full text: Technical Commission IV (B4)

   
ve and improve 
parameters are 
1, the initial value 
ral line should be 
tion will be non- 
? chips and initial 
llel processing of 
arch, which is the 
/ on single scene 
ying many scenes 
especially in the 
block adjustment 
1e future work for 
NT 
ligh Technology 
Sensing Satellite 
base Construction 
'y Assessment for 
N0.3,Mar.2009. 
esolution satellite 
1g algorithm and 
s,2004. 
te Geometry 
CD Line-scanner 
netry and Remote 
The Principle 
'ress,2002. 
gyi. Research on 
ge of HJ satellite 
Sensing and GIS 
PR 2009. 
trically Invariant 
pace Theory [J] 
2006. 
natching Points 
‚ Image Based on 
dics & Spatial 
008. 
justment. Wuhan 
tation of Hubei 
lel Programming 
2010. 
  
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B4, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
A PEDESTRIAN ACCESSIBLE POSITION EXTRACTION METHOD OF EXISTING 3D 
FILES FOR LARGE BUILDING EVACUATIONS 
Lei Niu®* , Xinli Ke”, Zhiwei Qiu? 
* Faculty of Surveying and Spatial Information, Henan University of Urban Construction, Postcode 467036, 
Pingdingshan, Henan, China, niuneilneo@hotmail.com 
® College of land management, Huazhong Agricultural University, Postcode 430079, Wuhan, Hubei, China, 
kexinliky@163.com 
Commission IV, WG IV/8 
KEY WORDS: 3D; Large Building; Evacuation; Pedestrian Data Extraction; Semi-automatic Solution 
ABSTRACT: 
As the emergency evacuation research in large building area draws more attention than ever before, it is natural to fast acquire the 
navigation information for this purpose. Current solution for extracting human accessible area from existing data files consumes a 
significant amount of resource and time. Thus a better solution is required. We propose a semi-automatic plan, which introduces a 
conceptual model to extract and organize the accessible data of 
arge building. This solution utilizes several spatial algorithms to 
extract detail traversing information from existing 3D building files and introduce spatial relationships to manage the extracted data. 
1. INTRODUCTION 
People need to face several challenging problems in modern 
society, such as natural disasters and artificial ones. Whatever 
the disaster is, we should find a fast responding plan to address 
the emergency caused by these disasters. Among this approach, 
the emergency responding in building-intensive area draws the 
attention of researchers, and the hotspot in this area is people 
evacuation in large building group. 
The people evacuation requires a delicate plan to meet the 
demand of directing people out of the emergency scene fast and 
safely, and this can only be achieved by executing evacuation 
simulation with real people and environment or with software- 
simulated people and environment. The former solution could 
generate more workable evacuation plan but cost more 
resources than the latter one. Therefore, the promising research 
trend is to choose the second way that uses the evacuation 
simulation program to mimic the emergency situation, produce 
and evaluate the evacuation plan. 
Under usual circumstance, the evacuation simulation program 
needs navigation data extracted from the real building with high 
accuracy and volume (figure 1); otherwise the simulation result 
cannot help researchers analyse the possible emergency 
situation. This means the supply of navigation data is crucial for 
evacuation simulation. 
X, 
$ 
iif A 
   
X 
\ = / 
Analyse 
  
; oy Sel 
Uilding ? 3 
Expert 
Structure Dat; e 
ücture Data Knowledge poser 
Figure l. Concept model of pedestrian data extraction from 
building structure data 
Researchers would face two challenges in order to acquire the 
Proper evacuation simulation data. The first challenge is to form 
1 Lua 
* . 
Correspondin g author. 
up a spatial relationship model to extract accessible data from 
existing building structure files, and the second challenge is to 
finish this extraction task for high volume data in a comparative 
short time period automatically than manually. 
These two challenges could only be overcome by utilizing 
proper methods. For example, introducing parallel computing 
technology could meet the computational demand of the high 
data volume of simulation data; while the complex spatial 
relationship model could only be constructed by improving 
existing models across several related disciplines. 
2. LTERATURE REVIEW 
Different researchers have taken various approaches to prepare 
raw navigation data for evacuation simulation. Lee and 
Zlatanova has proposed a solution to introduce the spatial 
topological relationships into the working procedure of 
extracting communication network from existed CAD files, 
which records the building structure(Lee and Zlatanova 2008). 
Li and He took a further step to prepare accessible data ready, 
and they combine the routing context information with the 
graph-extracted communication network(LI and HE 2008). 
This approach is followed by several researchers. Nagel has 
successfully produce some semantic information of building 
structure from un-interpreted data files, and this finding could 
also be easily transformed to extract communication data for 
emergency evacuation simulation(Nagel, Stadler et al. 2009). 
Furthermore, Boguslawski has introduced a boundary object 
called ‘cell’ to improve the extraction efficiency(Boguslawski, 
Gold et al. 2011). 
Besides the CAD extraction research, there is another approach 
discussing the 3D discretization of the accessible position for 
building environment. Bandi and Thalmann proposed a method 
of discretize the accessible plane for human navigation with 2D 
cells. Yuan and Schneider argue that a LEGO representation 
   
    
    
   
   
   
   
   
  
    
    
   
   
     
   
  
  
    
   
   
  
   
    
  
     
    
   
  
    
    
   
    
   
   
    
   
     
   
   
       
    
    
   
   
	        
Waiting...

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.