Dowman, Ian
4.2 Extracting the pixel value and writing the texture
The pixel value of a 3D grid point refers to the grey level of its projection on the image defined in the previous Section. In order to
define this value, the coordinates of the point on the image are computed. Then, using a simple interpolation the grey level associated
to the point is estimated. Once the image coordinates of the 3D point are computed, the texture value of the point is to be defined. For
this, the image is loaded into the memory by writing all its elements to an array with the same dimensions as those of the image. Each
element of this array is identified by a set of integers (i, j) which correspond to the computed y and x respectively. The grey level of the
point on the image, is the value of the array at ? — y and j = x.
Having obtained the pixel value of a 3D grid point, it is placed in the correct position which is defined by its indices (ie row and
column) on the grid. The process is repeated for all points of the 3D grid until the complete texture is produced. The task of extracting,
rectifying, and mosaicing image portions covering the face of interest, are carried out simultaneously in a pixel by pixel manner. The
result is a unique homogeneous rectified image of the building facade which is used as its texture. To actually make use of the texture,
it is mapped to the corresponding building face when the VRM is loaded. In other words, the building surface which is initially a blank
shaded face, is rendered with the texture used as the render material (Figure 6). The result is a geo-visual model, the components of
which are the geometric model and the textures obtained using the ATPT.
(a) Simple model (b) Texture (c) Textured model
Fig. 6: Building model before and after texture mapping.
5 ACASESTUDY
The area of this case study is one of the buildings of University College London (Figure 7) which is located at the west side of the UCL
quadrangle. As already mentioned, to produce the VRM of the building, a NFR is required which is enhanced with textures extracted
from the TID. The NFR defines the geometric structure of building while the TID contains the necessary data to extract the textures.
Fig. 7: A view of Chadwick Building.
As shown by Chapman ef al. (1994), using the registered terrestrial CCD images included in a TID the geometry of buildings can be
extracted accurately. Therefore, in this case the TID was used as the source to measure the coordinates of the building corners and
to form the textures. The result was the geometric model of the building which was rendered with the textures of its facades. The
textures were processed using the ATPT; for each face of the building, a texture was created, the data was structured using VRML for
visualisation. Figure 8 shows snapshots from the completed VRM from two different views.
184 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.
As
use
The
5.1
To «
the
sys
COO
plac
col
WIC
Fig
buil
wit
inte
Fig
c
pix
(A
are
urn
Fig
pix
on
ge