X-B4, 2012
le, the limitation
must be defined.
ion would appear,
lirectly from one
itation for people
limitation. Every
ur position must
people, otherwise
e of limitation is
th limitation, the
positions should
type of moving
of limitation is
ibouring position
c three types of
munication data
) the construction
r-space of large
or inner-building
' types of spatial
; next section, we
cd spatial objects
f candidates for
/. In detail, the
he inner-building
is determines the
and leads to the
e have evaluated
ose the cuboid,
disadvantages in
e cell representation
n from existing
cell form
x rectangle faces
, then each unit
Jakes an obstacle
. This is because
tion choices for
we six direction
After the setting of basic units, we can only use larger cuboid as
an index to represent a specific collection of basic units. And by
using the cuboid index, we could filter the unnecessary space
for a specific extraction operation out. F
Besides the introduction of basic unit and spatial index object,
we should also introduce triangles to represent the boundary of
the existing 3D objects for large buildings. The reason is that
the evacuation position extraction currently set the data source
to several common formats of 3D files using the triangle to
compose the building structure. Thus only after fully analysing
information kept by triangles could we finish the task of
evacuation communication data extraction.
312 Normalized Spatial Relationship for Inner-Building
Space
The triangles and cuboids play different roles in our normalized
spatial relationship model. The triangles in the existing data are
used to form up the skeleton of the building structure, and this
function is transferred to the boundary forming objects in our
model. These boundary objects could distinguish the inner-
space with outer-space of the building. The inner-space of the
building would possibly be involved in the evacuation
simulation, for they may be the communication position in the
next step. Nevertheless most of the outer-space will be filtered
out for being little value in evacuation simulation. Furthermore
we should also create a standard to differentiate the involved
inner-space from non-involved inner-space for a specific
evacuation research. We believe this standard could only be
defined by analysing the communication information from two
aspects.
The two aspects are whether current considering position is in
the possible inner-building communication space and whether
this position is accessible physically by normal people. The first
question could be answered by following this logic. The
triangles of the building representation successfully define the
inner-space and outer-space of the research building. We could
properly determine the accessibility of current position by fully
analysing the relationship between current position and all the
triangles belonged to the same building.
The second question could be solved by the establishment of
accessible position evaluation standards proposed in the former
section. This means we must check the moving length, height
and slope restriction to finally figure out whether the
considering position could be moved into from its neighbour
positions.
313 Manual Settings for Accessible Position Extraction
The accessible position extraction theory is the core of our
solution. And this theory mainly covers two aspects explained
in the former section and several key topics about other
Important details in the communication position extraction
process. First, we must clarify that the basic communication
unit in our research is the cuboid object with a fixed width,
length and height, which is normally a long 3D box with a
Square bottom. The fixed size of 3D box helps us format the
Whole area into a 3D array. This array could be well organized
the accessible information of the building part (figure 3).
Furthermore, the equal size of length and width is intended to
Tegularize every horizontal moving step length committed by
Simulated people.
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
Upside
Plane Slape
Figure 3. Evacuation area represented by regular accessible 3D
boxes
Second, the considering target (pedestrian) of the evacuation
simulation restrict the moving style to walking. This indicates
no long distance and cross-layer jumping is allowed. Thus one
building layer is divided into several layers formed by basic
units to prevent cross-layer moving.
Beyond the topic of basic unit and basic moving styles, there
still are many subjects not covered in this approach. They will
be supplemented in the future papers.
3.2 Key Algorithms for Evacuation Data Extraction
The key algorithms for evacuation data extraction are inner-
space determination algorithm for buildings, accessible position
analysis algorithm for buildings. The former algorithm is to
evaluate whether the current position is in the inner-space of the
specific building; while the latter algorithm evaluate whether
the inner-building position is an accessible position. The two
algorithms are combined to produce the result of accessible
communication position for specific large buildings.
3.2.1 Inner-space Determination
Researchers want to know one position is whether topologically
inside a building. This analysing operation is comparatively
complex. In this operation, we should not only decide whether
the position is in the convex boundary of the building, but also
evaluate whether the position is in the bounding area of the
triangles belonged to the building. This task could be simplified
by treating the position as a square unit. Therefore, we could
use the centre of mass evaluation method when we need to
determine whether this box object is inside the polyhedron
representing the building. In short words this means we only
need to consider the mass centre of the box object is in the
polyhedron or not. As is shown in figure 4, we could see the
belongingness for the small box unit is determined by the mass
centre position to the large polyhedron.