ame effect 
her digital 
>asurement 
ort on this 
[CS 
roperties of 
for image 
vive a first 
effects that 
se in terms 
. Although 
shows the 
very long 
lack target 
tefacts like 
lark current 
function of 
vas covered 
er was also 
ken with f- 
ne series of 
ied between 
— uaa: 
  
|, contrast 
e. It can be 
yr sensitivity 
respectively. 
> of the most 
illustrated in 
Jor corner) is 
identified as 
ve hard disk 
camera back. 
uality but, it 
image space 
(compare Hastedt et al. 2002). In order to avoid this effect it is 
strongly recommended to use a CompactFlash card instead of 
the MicroDrive device. 
  
  
Fig. 6: Effect of local temperature changes 
The Kodak Pro Back consists of an internal processor that is 
used for intrinsic image processing such as sharpening or color 
correction. It is not known how the implemented algorithms 
work. However, some effects can be shown by appropriate 
testing set-ups. One interesting software feature is the Long 
Exposure algorithm that corrects for effects due to the above 
mentioned dark current noise. The powerful algorithm 
generates images of higher radiometric quality but, the internal 
processor requires computing times of minimal 18 seconds. In 
addition, the processor is heated up during processing. After a 
five hours period of image acquisition the camera shut itself 
down due to overheating. 
4.2 Color processing 
One of the most interesting specifications of the Pro Back is the 
ability of generating high-quality color images. In order to 
receive first quantitative results a number of test images have 
been taken to check for sharpness and color reproduction. For 
this purpose test charts according to ISO/IEC 15775 (2A, 3B, 
4A) have been imaged under various exposure conditions. 
Additionally, a shaving brush and a spanner have been 
photographed in order to investigate resolution power and anti- 
blooming capabilities under realistic illumination and object 
conditions (Fig. 7). 
One simple test for color quality is the evaluation of a certified 
edge wedge for the three different color channels separately. 
Fig. 10 shows the resulting histograms for the RGB channels. 
The original grey wedge consists of 20 steps with a density 
range of 0.05 to 1.95 D. Using the integrated infra-red filter on 
top of the CCD sensor area up to 18 steps can be separated 
within the histogram of channel Blue and Green. In contrast, 
channel Red shows only 16 separated peaks where the lower 
(darker) regions can not be resolved sufficiently. This result 
shows that the color quality is poorer in the range between 
600nm and 700nm wavelength, yielding to a slightly worse 
image quality in dark object regions. It is obvious that the 
overall infra-red sensitivity of CCD arrays can not be controlled 
perfectly, especially for the Red channel. 
483. 
  
  
Fig. 7: Test set-up with color charts, resolution test patterns and 
real objects (complete image) 
  
Fig. 8: Grey wedge 
In order to investigate the sensor characteristics without infra- 
red filter a number of images have been taken under identical 
lighting conditions. As displayed in Fig. 11 the resulting 
histogram shows a lack of quality in the darker areas for all 
color channels. The dynamic range gets lower due to the infra- 
red sensitivity of the CCD sensor elements that exists even for 
color masked imaging arrays. As a consequence, the build-in 
infra-red filter should not be dismounted. Nevertheless, the 
optical characteristics of the filter plate (refraction, flatness) 
must be taken into account for photogrammetric camera 
calibration. 
4.3 Image quality 
  
Fig. 9: Test of image resolution for difficult object structures 
(1:1 subset of Fig. 7)