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 
Projection of PSS and MSS sensors on earth surface you can see 
on figure 5. Span width for PSS sensor (6 CCD matrixes) will 
be 23.3 km, for MSS sensor (4 CCD matrixes) span width will 
be 20.1 km 
Ei MSS - IR 
MSS - Red 
em PSS - 1-6 
pnm MSS - Green 
Figure 5. Projection of PSS and MSS sensors on earth surface. 
2.4. Source data modeling 
For developing technology source data were obtained by 
modeling sensors and shooting jf *Canopus-V" satellite 
[Nekrasov,2010]. Modeling images were done with the help of 
next technology [Nekrasov,2011]. High resolution image of 
"Resurs-DK-1" satellite was get for whole area of experimental 
testing area. Heights differens of testing area is about 400 m. 
We used digital elevation model (DEM), which was processed 
from SRTM (Shuttle Radar Topographic Mission) data. 
Orthoimage was processed using this hi-resolution image and 
DEM. GSD for orthoimage is 1 m, accuracy of external 
orientation is about 2.11 m. Orthoimage and DEM used as 
source data for creation of simulated images of *Canopus-V" 
PSS and MSS sensors. External orientation of simulated images 
is defined with the help of rational polynomial coefficients 
(RPC). RPC were calculated with errors of navigation system 
for simulation of real shooting conditions. 
2.5. Methods, algorithms 
RPC are based on next equations, connecting ground geodetic 
coordinates with image coordinates: 
Le Jip An hy) 
gi(9y An, hy) 
oo = Js Any) 
von df 
gs; (Ow. Aw. y) 
Nominator and denominator of this equations are polynomials 
of third degree: 
3 b dul 
2 du Ah 
3.13 
i=0 j=0 k=0 
Normalization of pixel and geodetic coordinates is done 
according next equations. Normalized values of coordinates will 
be not more than 1: 
1-0, 
In = — 
S, 
s-0, 
SN = ; tem——————— 
S e 
9-0, 
PN: = 
Se 
A03 
An = 
S; 
h-0, 
hy = ——— 
Sh 
Output data will contain polynomial coefficients fio. Dto Cile 
dii, and also normalizing parameters Oj, O,, Oy, Oy, Oy (Offset) 
and S,S,, S, S, S, (Scale). 
RPC calculation will be done according such algorithm: 
e Calculate external orientation parameters for image 
according rigorous sensor model and using control 
points (or we can use external orientation parameters 
calculated from navigation data); 
e Calculate for all image and heights evenly distributed 
control points (xg, yr, X, Y, Z) using rigorous sensor 
model; 
e Create equations and solve system with the help of 
least mean square method, for RPC determination. 
Tie points were searched in overlapped areas of CCD matrixes. 
First video data lines placed in focal plane to overlap each third 
frame of next matrix with first frame of previous matrix. Video 
data frame consists of 985 lines. Every next frame in MSS has 
overlap with previous frame in 57 lines. In PSS every next 
frame has overlap with previous frame in 80 lines. Between 
frames of different matrixes overlap is about 70 pixels. 
We use combination of area matching algorithms and SURF for 
tie points search. 
Block adjustment is done with the account of CCD matrixes in 
focal plane and using RPC. For block adjustment with tie points 
we use additional information about surface heights from digital 
elevation model or average height of earth surface. 
One of output products is synthesized coverage with sizes from 
23*10 to 23*20 km. This product can be used for increasing 
productivity of image processing. Whole pixel coverage for 
interesting area is done from adjusted block of images 
excluding overlapped areas. RPC calculated for such 
synthesized coverage. 
While moving along orbit arises overlap of frames in different 
spectral bands for MSS sensor “Canopus-V” satellite. We have 
possibility to get digital elevation model by combined 
133