4-5-4
G),, (fi, 5
Ci) 0 . (ft, K, 6 CO , 6 <j) , 6 K
Strapdown
INS
Mechanization
X,7, R(o) ,(p)^
Aerotrianguiation
estimation of
- INS errors
- System calibration
I A A I
GPS
Remote phase , > GPS data
doppler,
Master P seudoran 9 e > processing
==LJ
R(w . <p) , k
^GPS’ V GPS
Y* v
GPS ,V GPS
matcSina
Image data 9 > Coplanarity
> Offset correction <
A
Spatial offset
Exterior
> orientation
X, R(co , <p) , k
Ground control >
measurement
Collinearity
Figure 1 : Workflow of georeferencing.
To take the different error terms into account, the standard
collinearity equation
X 171 = Xg 1 + A R™ • x 9
(2)
where
coordinates of perspective centre
denoted in the mapping frame m
coordinates of object points given in
the mapping frame m
rotation matrix from image p to map
ping frame m
coordinates of image point given in
image frame p
scaling factor
has to be modified as follows:
X m = it™ + Rr • [A R™ • & + AX b am - a4 ps ] (3)
where
K= (
'5)
coordinates of the phase centre of
V ^0 )
the GPS antenna denoted in map
X m = [
< X\
ping frame m
Y
coordinates of object points given in
\
\ z )
the mapping frame m
( AXcam \
AYcam offset between INS and perspective
AZcam ) centre of imaging sensor given in
body frame 6
/ AXgps \
AXq PS = AV G ps offset between GPS antenna and
\ AZqps / perspective centre of imaging sen
sor given in body frame b
R™ = R™(u, <p,fi) rotation matrix from INS body 6 to
mapping frame m
r£ = R b (5u,6tp, 5k) rotation matrix from camera frame p
to INS body frame 6
coordinates of image point given in
image frame p
A
scaling factor
Equation 3 gives the complete mathematical model for the di
rect georeferencing of the image data. The different error states
mentioned in section 3.1 are integrated in the extended collinear
ity equation. Setting up this equation for each ground control
point or tie point allows the estimation of the unknown INS errors
(^0)y^o,^Oi^iji K i an< -l misalignment 6uj,6ip,6k between
the INS body b and image coordinate frame p. After estimating
the unknowns in the combined aerial triangulation, the results are
fed back to the mechanization to improve the starting values sig
nificantly. Using these corrected parameters the mechanization of
the INS data is repeated, and the improved exterior orientations
are used as input for a second adjustment process. Therefore, the
combined evaluation of GPS, INS and image data consists of two
major parts: The strap-down INS mechanization and the subse
quent aerotrianguiation for error estimation. The whole process is
performed iteratively until the final solution is reached.