ISPRS Commission III, Vol.34, Part 3A „Photogrammetric Computer Vision“, Graz, 2002
entity is either sunny or shadowed. Problems rise with respect
to the number of voxels multiplied by the computation time
used for energy flux balance and temperature determination for
a voxel. In a similar way, geometrical facets are heterogeneous
with respect to boundary conditions existing on these entities; if
computed on facet, surface temperature would not be physically
relevant.
Taking into account variations in time of all the interactions
differentiates the image synthesis in the infrared range from that
in the visible domain. The synthesis methods used for the
simulation of landscapes for short wavelengths do not need to
reproduce the recent past of the landscape as it is observed in
real infrared images. The radiosity method can be used in such a
case. In the thermal infrared range, it cannot, except for
considering the radiosity method for each voxel and each time-
step. An entity permitting to compute both temperatures at each
instant during the process and the radiance leaving the scene
has to be used.
In addition, a 3-D description of the landscape is necessary to
perform a simulation with a very high spatial resolution. In this
respect, objects forming the landscape interact each other.
Realistic image simulation should reproduce these physical
interactions between objects (Poglio ef al., 2001a). This adds to
the complexity of the simulation process.
3.3 The landscape representation
We defined an element as an entity that exhibits homogeneous
properties with respect to geometrical considerations, material
constitution, and the occurring physical phenomena at each
instant (Poglio et al., 2001b). To reduce as many as possible the
number of elements, and afterward the computation time,
element is defined as the largest entity included in a facet.
The element is a part of an object; it is a 3-D entity. Its external
surface is flat and oriented. The in-depth constitution of the
element is made of one or several layers of primary materials:
for instance, a wall can be made of two layers; construction
concrete and insulation. The element is homogeneous with
boundary conditions, both internal and external. Figure 1
illustrates the element.
To better understand the element, one may consider a facet; this
facet is partly shadowed at instant /. These portions of the facet
are a preliminary set of elements, which will be subsequently
subdivided by taking into account other phenomena and other
instants.
The mesh supporting objects constituting the landscape is made
of the union of elements for the period of simulation of
temperature. Figure 2 illustrates the representation of the
landscape. The landscape is divided into objects, objects into
facets, facets into elements. In each subdivision, properties and
characteristics are inherited from the previous hierarchical level
and are added to perform an element-based representation of the
landscape.
A - 242
Homogeneous flux
balance at each point
on the external surface
Multi layered |
constitution in
the in-depth
dimension
Same boundary conditions
on internal surface
Figure 1: illustration of the multi-layered in-depth constitution
of an element and the homogeneity with respect to
boundary conditions on both the internal and
external surface.
| Landscape |
T
v
Facet k-1 and
3-D description
Objectk |
Facet k-p and
3-D description
Element k-p-1
Y
Facet Æ-2 and
3-D description
Element &-2-5
|
Element &-/-/
Element &-2-/
Figure 2: illustration of the landscape element representation.
4. ARCHITECTURE OF THE SIMULATOR OSIRIS
The simulator OSIRIS is divided into four primary simulators,
each of them dealing with a particular sub-part of the simulation
(Poglio et al., 2001c). These four primary simulators, called
(S0) to (S3), operate successively.
The main technical originality of the simulator OSIRIS is the
pre-processing simulator, (SO), generating the elements. The
simulator (S0) operates with the following scheme (figure 3).
The shadow maps are made and combined with the facet
description of the scene to obtain the element-based
representation. In addition, the radiative and conductive
surroundings, the wind velocity for each element and the form-
factor matrix are computed.