)is not a 
h a very 
ncreased 
:dundant 
generate 
rom the 
9 images 
  
analyzes 
pixel on 
el on the 
ed by its 
e image 
> highly 
sidelap), 
pixel on 
rocessed 
/ales per 
ust and 
itcher of 
oints per 
airborne 
  
Figure 4. High resolution 3D point cloud generated 
by the dense matcher module of UltraMap 
Figure 4 shows the point cloud generated by UltraMap 
based on a set of overlapping UltraCam images. The point 
density is around 300 points per square meter. For better 
visualization only a quarter of all points are shown. 
The achievable height accuracy of the point cloud is better 
than the GSD, thus a 10cm imagery leads to <10cm height 
accuracy of the resulting digital surface model. The quality 
of the automated dense matching process depends 
significantly on the camera and the structure of the terrain. 
Geometric stability and radiometric dynamic of the camera 
have a direct impact on the matching results. 
3. DSM and DTM 
Once the DSM has been processed out of the images, the 
DSM and the original image can be combined in a true- 
ortho projection which results on a so called true-ortho 
image. This is an image with no perspective view as each 
pixel of the image has been (true-) ortho rectified by a 
surface model and not by a terrain model like the standard 
ortho image. True-ortho images have a significant benefit 
in some applications as they do not have any leaning 
objects. 
Figure 5 shows the shaded relief of a DSM which has been 
generated from the point cloud. Remarkable is the high 
level of detail and also the very sharp building edges. The 
white spot at the lower left corner is the church tower 
shown in detail in figure 4. 
Figure 5. DSM generated from the point cloud by the 
UltraMap dense matcher module 
The next step after the DSM generation is the processing of 
a DTM. The DTM will be processed out of the DSM using 
a “Winston-Salem” algorithm developed by Microsoft 
which utilizes computer vision based classification for the 
terrain filtering. Figure 6 shows the DSM (left) and the 
filtered DTM (right). 
  
Figure 6. High-resolution DSM (left) and DTM 
(right) processed using automated dense matching 
with UltraCam images and UltraMap processing 
software 
4. DSMOrtho and DTMOrtho 
An additional new UltraMap 3.0 module then combines the 
image data and the DSM or DTM and generates 
automatically either a DSMOrtho or a DTMOrtho image 
including automated seamline generation. Figure 7 shows a 
DSMOrtho image, figure 8 a DTMOrtho image. 
    
    
E M Xm: S 
Figure 7. High resolution DSMOrtho, generated 
automatically by UltraMap 
scone eem ME ee ffomsmsne