Figure 1: Test assembly at the first camera location. 
HRSC is enclosed in a vacuum tank in order to simulate 
conditions in space. WAOSS (Wide Angle 
Optoelectronic Stereo Scanner) and a video camera are 
attached on top of HRSC. 
expected in Mars orbit. The experiment produced about 50 
panoramic views with maximum sizes of up to 150,000 
lines, each line containing up to 5,184 active pixels. High 
rate DCT (Discrete Cosine Transformation) data 
compression was applied during imaging with compression 
ratios of up to 40 depending on a selected quality factor, the 
selected code tables, and the scenery. Typically, 
compression ratios were selected to be small for the color 
channels to retain optimum radiometric accuracy. The 
images (Fig. 2) cover a large part of Lake Constance and the 
  
  
  
  
  
  
verification of telemetry data, sorting of house-keeping and 
sensor data, data decompression, computing of time tags for 
each recorded image line, and correcting for flat-field effects 
(Oberst et al., 1994). 
In order to analyze the geometric properties of the HRSC 
imagery, the disparity of respective image pairs were 
analyzed by digital image matching. Two different area- 
based matching programs were available: (1) Gotcha, 
developed at UCL (University College London) (Day et al., 
1992); and (2) CLTMATCH2, developed by TUB (Technical 
University of Berlin) (Oberst et al. 1996). Since 
CLTMATCH2 is a precursor program to the software 
designed to be used in the HRSC/WAOSS ground data 
processing, the images acquired during ET3 provide a good 
opportunity to test this software. 
3. RADIOMETRY 
3.1. General Image Quality 
Sporadic consistency checks on the large amount of 
available data show that the general appearance of image 
data is excellent. The images are well focussed, and the 
exposure times were selected properly. However, due to the 
wide dynamic range in brightness of the scenery, small parts 
of the image are saturated or underexposed. There are no 
missing or extra lines. Pixel binning and changes in 
exposure time perform as required acording to camera 
commands. The excellent MTF (Modular Transfer Function) 
of the HRSC yields sub-pixel resolution in regions with 
high image contrast (Fig. 3). 
  
   
  
  
  
   
  
  
  
  
   
  
  
  
  
  
  
  
   
  
  
  
  
  
  
  
  
Swiss Alps. S 
3.2. Brightness Comparison > 
First location Second location 5 
North 47° 40.25' 47° 36.66' To verify the radiometric performance of the HRSC, image g 
Bast 9° 23,54" 9° 35.96" brightness of the five panchromatic CCD sensors have been & 
Altitude: 425 m 434m compared. Histograms of DN for a subscene of an ET3 image 
(Fig. 4) demonstrate that variations in brightness between 
: ; : the images of the single CCD sensors are relatively minor 
Table Geographic «oordindles of Camera and the overall shape of the histograms is identical. The 
. fact that images obtained by the off-nadir channels appear 
darker than the nadir channel images (see detail of Fig. 4) 
can well be explained by the metric properties of the camera Fig 
2. DATA PROCESSING and by the imaging geometry (see 4.2). The result indicates ster 
: : : : that the radiometric performance of the HRSC is extremely 
All images were systematically processed which included accurate even for raw images. 33 
Dur 
NT EU A OKT ima 
isa —— —— Stereo 1 sen: 
ET. Infrared visi 
strij 
Photometry 1 
Green 
Nadir -———» Nadir 5 
o 
Blue ——-— 2 
Photometry2 i 7m —> 3 
Red -— ——- 5 
Siereo 2 «7777774 
Figure 2: Example of ET3 Imaging Sequence viewing Lake Constance and the Swiss Alps. The single channels are all 
individually contrast enhanced, resampled to a common macropixel format, and arranged in a way that identical objects Fig 
observed by different channels appear vertically aligned. Note that the direction of the rotation was from right to left. (nac 
350 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996