§
b
a
t
T
}
s
H
Hu
1B
i
i
d
1
"
H
$
H
i
t
A
i
i
ng
i
ih
|
1
ti
|
2. MATHEMATICAL MODELS
2.1 Space Resection
Photogrammetric collinearity conditions used for
SPOT panchromatic imagery with linear CCD
arrays are:
x;+a0,+al,s;+a2,57 mu
Aux = Xos) t 812; (Y. = Yoz) +H a13,(Z; = Zoj)
In Eq.(2) d is the separation between two neigh-
boring sensor stations k and k + 1; station j lies
between k and k + 1, and j is away from k in s
units. For space resection, Eqs.(1,2) serve as our
functional model in least squares adjustments.
Based on weighted ground control points, we es-
timate mainly the position and attitude param-
eters .. Xo. > Yon: 207 3 ks Ok Kg... and the self-
a31, (Xi = Xo) d azz; (Y; "T yi) + ass, (Zi = Zoi)
(1a)
az, (X: — Xo,) +422, (1H: — Yo; ) + tas, (Zi — Ze, )
yi+a0y+al,y;+a2,y? = —C
where c : camera constant;
X;,Y; : image coordinates of point : and z; = 0;
Sj : Strip coordinate in units of length or time;
Xo; Yo; Zo; : time-dependent position param-
eters at sensor station 7 when point ¢ is im-
aged;
411,;...433, : elements of an orthogonal matrix;
they are functions of time-dependent atti-
tude parameters w;, $;, &j;
Xi, Yi, Zi : object-space coordinates of point 7;
a0,...a2, : additional self-calibrating parame-
ters.
In order to be practical for implementation, the
time-dependent parameters of exterior orienta-
tion Xo; » Yo; » 70; » Wj, Dj, K; are described by piece-
wise continuous linear models:
X,, 2 (1— s/d)X,, + (s/d)X
Ok+1
(1 — 8/d)Yo, + (S/d) Yo,
(1 = s/d) Zo, + (s/d)Zo,,,
(1 = s/d)wr + (s/d)wr+1 (2)
(1 — s/d)®x + (s/d)®x+1
(
1 — s/d)kr + (s/d)Ki41
Y
Z
wj
bj
Kj
646
(Yı — Yo,) + 433, (Zi — Zo,) (1b)
calibrating parameters.
2.2 Integrated Approach to Image Matching and
3D Positioning
The method of least squares image matching for
stereo images can be written as
Vy — g'(zj,y;) — ro — 1g" (27, y; Pg: (3a)
in which g', g" : gray-value functions evaluated
at two corresponding image points 77, y; and
a"! yl;
254429)
To9,T1 : radiometric additive and multiplicative
parameters;
vo Pol : gray-value residual and its associated
4 i
weight.
Logically and in a straitforward manner, the
collinearity conditions Eqs.(1 with 2) can be used
for pairs of image coordinates in Eq.(3a); and in
vector notation, we get
vu = g'(Zi p^ a) - ro - rig" (Zi, p^, a)i py:
(3b)
where p/, p" : vectors for position and attitude
parameters along single- and double-prime
orbital paths, respectively;
a : vector for self-calibrating parameters.