International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
  
Filling the remaining part of the DDTM (ground), with 
interpolation based on points of group n. 2 from which the 
points inside the areas processed in the previous step will 
be removed. In this way the interpoled surface also follows 
the ground even when there are bridges and similar 
infrastructures. 
4. Binary format writing of the DDTM file (cach height is an 
integer number). A file is also compiled, that contains a 
description of the number of columns and rows, the 
reference system, the step, the multiplying factor to use the 
AccOrtho software or any other commercial image 
elaboration software (ENVI, e.g.). 
In short, the interpolation techniques implemented in the 
software are: Nearest Neighbor, Mean Square Plane, Bicubic 
and Bilinear Spline [Brovelli, 2001]. 
An example of the output result of the GeneDDTM software 
can be seen in fig. 6. The first part (a) contains a portion of 3D 
base map at 1:1000 scale of the city of Turin (map sheet n° 
112). The second part (b) contains a digital grey scale image of 
the corresponding DDTM (with pixel size of 20x20 cm“, which 
is suitable for a 1:2000 scale orthophoto), generated with 
GeneDDTM in a processing time of about 30 minutes (for the 
entire 2300x3400 node sheet, using a bilinear spline) on a 
standard Pentium 4 1.5 GHz PC — 512 Mb Ram. 
LI 
  
Figure 6: a result obtained with the Gene DDTM software 
4. HOW TO PRODUCE AN ACCURATE ORTOPHOTO 
In 1996 Amhar and Ecker proposed an original solution for the 
generation of a true orthophoto. The procedure, devoted to the 
production of orthophotos in urban areas, used a DSM that was 
managed by a relational database. All the images were classified 
as terrain or building surfaces and the orthopoto was generated 
in separate phases: first the terrain then the roofs. The results of 
these treatments were then merged into a single digital 
orthophoto. Hidden areas were eliminated through 
superimposition of the orthophoto generated from other images. 
The proposed solution tries to simplify this approach. The input 
data for the generation of a true orthophoto are: a DDTM as 
previously described, and a series of oriented images containing 
the radiometric description of all the points to be orthoprojected. 
The aims of the procedure are: to maintain complete automation 
so as to guarantee the same performances as a traditional 
orthoprojection software and to avoid the previously highlighted 
problems. 
Let us consider the object in figure 7. In perspective images, 
higher points correspond to lower points, therefore the 
procedure must run from the highest to the lowest point. 
The procedure starts from point R. The best recording of the 
grey value of this point can be found in the image which has the 
projection centre nearest to the point itself (image I1). In order 
to avoid the duplication of the images (as see in fig. 2), this 
pixel should be inhibited: for this reason a "flag image" is 
created where each pixel records the height used for the 
540 
orthoprojection of the corresponding pixel on the original 
image. Point R has also been recorded in I2 and, for the same 
reason, the pixel that represents point R on I2 should also be 
inhibited. 
Qp Or Qr Qs 
Flag Images M. iei ene 
Images { | 12 
   
  
  
  
Figure 7: Ortoprojection procedure scheme 
The procedure orthoprojects point S with the same criteria 
(point S will only be recorded in Il). When the procedure 
orthoprojects point P, it finds the pixel on Il that was used 
before for point R. The flag image inhibits a second use of this 
pixel, because the height recorded on it is higher than the height 
of point P. Then the procedure looks for the grey (or colour) 
value in I2. The pixel is not inhibited and the orthoprojetion of 
point P is possible. When the procedure orthoprojects point Q, 
the first attempt is to use the corresponding pixel on I1, but this 
pixel has been used for point S and the "flag image" then 
inhibits the radiometric value reading. The second attempt is to 
use the corresponding pixel on 12, but also this pixel has been 
inhibited because it contains the grey (or colour) value of point 
R. In this case, no more images being available, the 
orthoprojection of point Q cannot be defined. This simple 
example describes all the possible cases of a true orthophoto 
projection. 
5. ACCORTHO SOFTWARE 
The procedure that is described in the previous section was 
implemented in a specific software called ACCORTHO 
(ACCurate ORTHOprojection). 
The input data consist of a regular DDTM, generated as 
previously described. 
The software works in two separate steps. In the first, it selects 
and prepares the data. In particular it: 
* calculates the heights of each pixel of the output image (a 
true orthophoto) and orders the pixels according to 
decreasing heights; 
e extracts the portions of the digital images involved in the 
orthoprojection; 
e prepares an index of the images in order to find the 
radiometric value to use whenever possible. The images 
are ordered on the basis of the distance between the 
projection centre and the considered pixel; 
® generates an empty flag image for each input image. 
The second step of the procedure puts the process described in 
the previous section into practise. Fig. 8 shows the flow chart of 
the basic functions. 
The functioning of the AccOrtho software has remained 
unchanged since it was first released [Dequal et al., 2001]. The 
evolutions that have been implemented in the last year concern 
some optimization and operative problems, and they have been 
Int. 
ma 
Th 
The 
pos 
OS