nbul 2004 
nner lens 
sor CED 
Im 
the same 
levice. In 
rted with 
ng flight 
led. Now 
portation, 
from the 
oduct . 
han film, 
combined 
ore more 
nation of 
ssing and 
ugh the 
stem 
lements 
ath two 
Light 
transfer 
physical 
the day, 
and the 
y value 
. Herein 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol XXXV, Part B2. Istanbul 2004 
  
  
23.2 (Geometrical correction of the lens 
Optical lens systems have geometrical errors, like distortion and 
astigmatism While astigmatism can be corrected to a minimum. 
distortion always has to be considered. 
Therefore the lens systems have to be calibrated. In case of film 
based cameras this calibration has to be repeated in a certain 
time frame e.g. every two years. This calibration data set is 
used to correct the images after the scanning process. 
In case of a digital camera, like Z/1 imaging’s DMC, the 
geometrical correction is applied during the post processing 
step, where the images from cight camera heads are combined 
to one large image. 
The accuracy can be tested on special test fields with known 
position of test objects. 
23.3 Radiometric correction of the lens systems 
Because of mostly used wide angle lenses in photogrammetry a 
relatively high apodisation can be detected. For film based 
cameras special grey filters are used to compensate this effect. 
This is the only possibility to do it, because of the relatively 
low dynamic range of film. 
With digital imaging a reference image can be exposed and 
stored in order to correct the image digitally. This possibility 
can be used based on a wider dynamical range of CCD 
clements. 
In combination with apodisation correction the pixel sensitivity 
correction is done, which is mentioned later. 
2.3.4 Resolution of the lens 
On film based cameras and scanners two different lens systems 
are used during image production. 
The first lens system is used in the camera itself, the second 
within the scanner. For both systems resolution criteria have to 
be fulfilled. 
The pixel size of the sensor determines the edge frequency of 
the so called MTF curve. The variable of this curve is the 
spatial frequency in line-pairs/mm. It determines the number of 
line pairs (black/white), which can be resolved in the image. 
The highest value fits with the pixel raster, when the change of 
white and black lines in the test grid have the same distance as 
the pixels of the sensor. So the edge frequency (EF) is defined 
as: 
EF = 500 / PX where PX is the pixel size in um 
For example a CCD sensor with I2um pixel size has to be 
combined with lenses resolving about 42 LP/mm.. 
In case of digital cameras these criteria have to be fulfilled by 
the camera lens only. 
  
247 
2.3.5 Compensation of sensors effects and defects 
No technical sensor array, either line sensor or two dimensional 
array, has a totally identical sensitivity for each sensor element. 
Film has the same effect, but because of the change of film 
material from picture to picture, this effect cannot be 
compensated. It is a statistical effect, called grain noise, that 
depends most of all on the material characteristics of the used 
film. The conditions of the chemical development process are 
important too. 
A compensation can be done and must be done for digital 
sensors. 
It is nearly the same process for scanners and digital cameras. A 
reference image of an object with totally homogenous light 
density distribution is stored and produces a data set for image 
correction. 
This correction data set has to be delivered with each digital 
camera or with each scanner. 
Within this data set the different sensor errors, like hot spots, 
dark pixel errors and columns defects are registered and 
compensated. 
This calibration process has to be repeated in case of the 
scanner in certain time frames. 
In case of digital cameras there is no experience until now, if 
this radiometric calibration has to be repeated. 
For our opinion it has to be done only once during production. 
This statement has to be proven within the next years, when 
cameras work out in the field. 
2.3.6 A/D Conversion 
This step influences the radiometric accuracy only. Here the 
producer of the camera or scanner has to guarantee his 
specifications. 
Scanners should have a dynamical range of 10 bit minimum. 
Digital cameras usually work with a dynamical range 12 bit. 
2.3.7 Combination of multiple camera heads 
In case of digital cameras more than one camera heads are 
combined, using identical cameras to increase the field of view 
and cameras for different spectral ranges. 
Cameras which cover the whole spectral range produce 
panchromatic images: These images more ore less determine 
the spatial resolution of the overall system. 
As an overlay for colour information called multi - spectral 
cameras are used. Depending on the camera system these 
colour cameras use less resolution than panchromatic cameras. 
The relation between pixel size in panchromatic images and 
pixel size of colour overlay images should not exceed 1:5.