primary
Complete maps Lire mapped
Revised
| maps
Upgraded
— key-items
Semantic Zt eributes
Thematic
OUTPUT 2
I content —— :ey-items
Metris ——lattributes
Digital
—— Presentation Graphic j specifications
Textual
: Semantic eps :
otre ; specifications
Metric
Fig. 4: Description scheme for the output.
The link between the description of input and output is established by
the specifications for the process and intermediate product.
5. AMP procedure
A suitable way of modelling the process is by a matrix, representing the
domains versus the process stages (figure 5). The operation begins in the
preparation stage by verifying terrain relief. Then it sequences along
the path through both domains, as indicated in figure 5 by arrows. The
nucleus of AMP represents the geometric transformations, especially the
"image-to-DTM surface” transformation. On-line systems perform these
transformations in real-time.
In the following section,
tranformtion.
attention is focused on the image-to-model
6. Image-to-model transformation
To this end, any image with known geometry can be used, provided that the
corresponding DTM data are available. Corrections for determistic errors
can be included.
The computation comprises two stages:
— External orientation (or determining the rotation matrix), which re-
stores the image rays in space;
- Ray intersection with the DTM surface.
The second stage is usually iterative, by applying a feedforeward (RADC)
or a feedback (ITC) process.
— The "Photo-Carto Interface System"
Development Center (RADC) begins the
developed by the Rome Air
process with a Z-plane, which is
above the highest terrain point in the
Then Z is incremented by dZ in the "top
tersects the DTM surface. For each
intersection point are caluclated. Then
image (figure 6).
down" sequence until the ray in-
Zi-plane, the Xi, Yi of the
the corresponding terrain height
Hi is determined by interpolation from the DTM data. The sign of the dis-
crepancy /Zi-Zi-Hi indicates whether the point i on the ray is above or
below the DTM surface. Thus, when 4Z; becomes negative, the iteration can
be terminated. For computation of the adjacent points, the Z-level of the
32