International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B4, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
processing of multispectral data. Main feature of such 
processing is in point correlation from different spectral bands 
data. One more feature of such data is low ratio B/H. And we 
must estimate possible accuracy of such processing. 
3. Experimental results 
Calculating RPC for PSS CCD matrixes, accuracy (RMS) of 
coefficient calculation for testing area is from 0.000693 to 
0.001306 pixels for normalized pixel coordinates. 
We use correlation coefficient not less than 0.9 for searching tie 
points. Bad tie points excludes in block adjustment of CCD 
matrixes for subpixel accuracy of block adjustment. For testing 
area we get block adjustment accuracy from 0.1 pixel to 0.8 
pixel (RMS) depending from block size. Georeferencing 
accuracy for testing area without control points is defined by 
navigation system accuracy and in simulation was set for 
position — 10 m (RMS) and for orientation angles 10 arc 
seconds (RMS) it will be equal to 80-100 m (RMS) on earth 
surface. Using additional control points radically increases 
block georeference accuracy (see figure 6). One control point 
increases georeference accuracy to 18 m (RMS), three control 
points decreases error to 4.3 m (RMS), in such case we have 
good block adjustment with tie points along and across track. 
  
Accuracy (RMS), m 
  
points 
  
  
  
Figure. 6. Georeference accuracy of CCD matrixes block 
External orientation accuracy for synthesized coverage with 7 
control points is 3.5m. 
Relative orientation accuracy of stereopair from multispectral 
images for testing area: 
RMS: 0.464 
Average: 0.397 
MAX: 0.834 
Points quantity: 30 
External orientation accuracy of stereopair with 4 control points 
is equal to 26.7 m (RMS). 
External orientation accuracy of stereopair with 7 control points 
is equal to 17.2 m (RMS). 
We compared source digital elevation model (from SRTM data) 
which was used for simulated images with calculated DEM.to 
Height errors were: 
RMS: 21.2 M, 
MAX: 32.3M. 
4. Discussion 
Results of this work may be used as guideline in using of 
“Canopus-V” and “BKA” data. Meanwhile we must remember 
that this results were get from simulated data. Simulation must 
be obligatory in designing of new remote sensing satellites and 
simultaneously, processing technologies of data from this 
satellites. 
5. Conclusions 
Results of simulation and accuracy estimation shows, that 
panchromatic sensor data of “Canopus-V” satellite may be 
recommended to modify maps for the scale 1:25000 
Multispectral sensor MSS data may be used to create and 
modify maps for scale 1:100000. Using of adjusted block of 
CCD matrixes allow to use “Canopus-V” sensor features with 
maximum accuracy. Synthesized coverage with RPC 
georeference allow to increase productivity of image 
processing. We demonstrated possibility of DEM generation in 
this work. But because of small ratio B/H it accuracy is not very 
high, meanwhile it may be recommended for using in 
processing for scales 1:100000 and less. 
6. Acknowledgements 
We would like to thank Nikonov Oleg - head of remote sensing 
department of VNIIEM corporation for his active support of this 
work. 
References 
Vladimirov A.V., Salikhov R.S., Sennik N.A., Zolotoi S.A., 
2008. Space system for operative monitoring of technogenic 
and natural emergency situations on base of space complex 
“Canopus-V” and Belorussia space satellite. Electrician 
problems, 105(1), pp. 49-57. 
Nekrasov V.V., Makusheva E.V., 2010. Development of 
dynamic geometrical model of shooting of optoelectronic 
sensors for perspective space complexes “Canopus-V” type. 
Electrician problems, 119(5), pp. 25-30. 
Nekrasov V.V., Kravtsova E.V., 2011. Processing technology in 
DPS PHOTOMOD images of perspective satellite “Canopus- 
V". Geoprofi, 5.2011, cc. 15-18. 
134