International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B3, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
A slight distortion of the internal model of the camera can also 
be seen in the spatial distribution of the height errors (Figure 2). 
height error + 
280 
‘ters 
nb of points 
= 
3 
180 r 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
     
  
  
  
  
  
  
  
  
  
-50 
Figure 2 Polynomial approximation of the height errors 
In spite of these sparse artefacts, the overall quality of the 
elevation surface is not very far from what we usually extract We m 
from airborne imagery and we can even expect an improvement T = using ? 
of the accuracy with a better knowledge of the camera geometry DANN EER SEE stereo 
after the completion of the commissioning phase of the satellite. Figure 4 Melbourne ground control points pas ( 
the inc 
5.3 "Melbourne All pairs with a stereo angle of (approximately) 15? have been error w 
processed. In a second time, we processed also the pairs with a measur 
A sequence of 17 images was acquired from -49.6° to 48.7° 10° stereo angle, from -21.7? to 20.7". situatio 
along the track, evenly distributed every 5 to 7 degrees. The tables below summarize the individual performance of each clearly 
stereo pair, in terms of completeness and accuracy. brings 
narrow 
Incidences Nb. Points Bias RMS LE90 higher, 
-49.6/-28.4 141 0.44 1.11 1.24 obtaine 
-42.3/-21.7 207 0.05 0.73 1.14 
-35.2/-16.0 244 0.18 0.83 0.95 Incid 
-28.4/-11.0 264 0.28 0.69 1.01 -16.0 
-21.7/-5.8 278 0.45 0.87 1.24 -11.0 
-16.0/-0.3 285 0.26 0.66 1.08 -21.7 
-11.0/5.1 291 0.37 0.75 1,12 -16.0 
-5.8/10.4 285 0.23 0.73 0.99 -28.4 
-0.3/15.8 277 0.24 0.58 0.97 -21.7 
5.120 7 271 0.29 0.96 0.99 
10.4/27.4 259 0.22 0.65 1.01 
15.8/34.3 238 0.22 0.72 1.08 The be 
20.7/41.4 220 0.33 0.81 1.06 obvious 
27.4/48.7 203 0.26 0.90 1.14 defined 
Table 1 Wide angle stereo pairs (base/height = 0.26-0.36) the per! 
This co 
Incidences Nb. Points Bias RMS LE90 environ 
-21.7/-11.0 270 0.52 0.93 1.52 everywl 
-16.0/-5.8 277 0.51 0.97 1.52 fly only 
-11.0/-0.3 289 | 049] 992] 138 the exac 
-5.8/5.1 291 0.42 0.78 1.24 
P : -0.3/10.4 202 0.39 0.89 1.32 
e NT fh a = 5.1/15.8 278 0.38 1.01 1.34 We ha 
Figure 3 Melbourne sequence samples 104/207 273 "ET [17 1.47 We DN 
A set of 295 check points have been manually chosen on the Table 2 Narrow angle stereo pairs (base/height — 0.18) elevatio 
reference DEM, partly on the ground and partly on the buildings ; satellite 
roofs (Figure 4), checking that no change occurs between the Those results underline clearly two phenomena: ; efficient 
airborne survey and the Pleiades acquisitions (the west part, ® The completeness of the Measures 13 highly dependent sequenc 
near the stadium is obsolete in the reference data). on the incidence angles (Figure 5). an open 
stereo p 
wide ste