The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Voi. XXXVII. Part B4. Beijing 2008 
374 
Consequently many attempts are being made to generate so 
called “True Orthoimages” as highest quality orthoimages. 
Related literature provides us with many interesting approaches 
for this purpose, but all require a detailed three-dimensional 
description of man-made objects in vector format, which is very 
difficult to achieve. 
However, after the correction of the displacements, the 
remaining empty areas, so called gaps, must be filled up with 
image data from a corresponding imagery. The result of the 
described process is an orthoimage in high quality, a "True 
Orthoimage", which shows all objects in their correct ground 
position. The process of true orthoimage generation, the 
geometric conditions and new improvements are published in 
the literature, e.g. Habib et al. (2007), Kraus (2004), Mayr 
(2002). 
Recent developments of optoelectronic line-scanning cameras 
allow an entirely new approach for orthoimage generation. The 
approach is based on the particular image geometry of the new 
camera systems. The next section demonstrates the specific 
feature of the image geometry and the background of the new 
approach 
3. OPTOELECTRONIC LINE-SCANNING CAMERAS 
AND THE NEW APPROACH 
In order to overcome the mentioned difficulties, a new approach 
for the generation of true orthoimages has been developed and 
presented for the first time by Albertz and Wolf (2004). The 
most interesting fact is, that this method does not require height 
information for true orthoimage generation. This is made 
possible through the development of optoelectronic line 
scanning cameras. Such cameras, operating in the so-called 
pushbroom mode, consist mostly of three or more CCD-lines, 
which acquire image data from the terrain surface by the 
continuous forward motion of the camera system and constant 
reading of the CCD-signals. This method follows the three-line 
concept introduced by Hofmann (1982). In theses cameras one 
sensor line is facing to the nadir and is adjusted perpendicular 
to the flight track. Thus the data acquired by this channel show 
mixed geometrical properties. In flight direction the terrain is 
imaged in parallel projection and across to the flight direction in 
central perspective (Figure 1). Now, if ideal flight conditions 
are assumed, the displacements of objects due to its height 
occur only across the flight direction. This means that the same 
objects show no displacements along the flight direction and are 
located in their correct ground position. The most important 
aspect is that this result is independent from the height of the 
objects. These unique geometrical properties have never been 
used in the traditional generation of orthoimages. 
Figure 1 shows a digital three-line scanner schematically. The 
overflown terrain will be imaged by the three sensor lines a, b 
and c, in the focal plane of the camera system. Under ideal 
flight conditions, the nadir-looking line b observes a differential 
line g of the terrain. Assuming a uniform forward motion along 
the flight line F and a constant recording rate, an image strip 
will be recorded showing the terrain in mixed projection. Relief 
displacements occur only along the line, so that the object H is 
leaned out within the line. But, a true ground coordinate exists 
in flight direction independent from the height of the ob 
ject. 
Figure 1. The principle of a three-line scanning camera 
The new approach for the generation of true orthoimages makes 
use of this particular image geometry. However, this requires 
that the same terrain must be imaged twice, with the condition 
that the second flight line is oriented perpendicular to the first 
one. Consequently the direction, which was previously imaged 
in central projection, shows parallel projection in the second 
flight line, and thus provides the second ground coordinate. 
Therefore, the basic idea of the new approach is to combine the 
two strips to produce data in parallel projection, i.e. true 
orthoimages, in one coordinate system. The principle is 
schematically illustrated in Figure 2. 
Figure 2. The combination of data from two flight lines 
to form a true orthoimage 
The top level shows the object H with a specific height above 
the reference plane. The surface of the object is imaged at the 
position by flight line FI and displaced across (1). The same 
surface is imaged towards the other coordinate direction by 
flight F2 and even displaced across (2). This means, correct 
ground coordinates exist in the flight directions. By combining 
the given coordinates, the image pixel or ground segment H can 
directly be mapped into the matrix of the true orthoimage plane 
(TO). 
It is evident that the successful application of this approach 
depends on two conditions: 
1. The geometrical assumptions must be fulfilled, i.e. the 
imaging plane must be vertical, and the flight lines must be