id camera atti-
elow.
(6
(6)
ge Iorth, and
h minimize the
Kazuo Oda
e x, y, Eall, i L = lorth, X. y E,, I, - Iorth, XY, E,, I, (7)
where E, and E, are absolute orientation parameters of stereo image 7 ; and L5:
t t 1 t
-- |/ Í = = lf f =
Eo roli 1 1X Y, Zj eT E
(8)
t
and Eall = [, 'e)
We can define evaluation function KEall for least square optimization as below:
n
Eall = ex, y; Eall I, I, ? (9)
i
where n is the number of points on DSM which are seen on both of the stereo images.
2.3 Least-Square Optimization
We have adopted Gauss-Newton method for non-linear least square optimization. This method revises target parameters
Eall with Eall:
Eall+ Eall Eall (10)
Eall can be computed by the following equation:
=1
Eal --'A A ‘À € (11)
where
ft
€ = le, e ej (12)
t
t t t
A= e: e en (13)
Eall Eall Eall
This computation is repeated until the evaluation function — Eall converges on the minimum value.
e;
. . . 1 . .
Partial derivative Eall © be computed by following equation:
a
e
= ei e; = I; Pbhti I Ppp
Eall VE. "TP. CE. COP. UE
E, E, Pont1 E, Poy E,
(14)
where
P phil "phil "phil "phil "phil "phi
I, I I A phtl. | 1 1 : 7 '" 7
= 1 / E
Pont1 E] Y» 1 phi phi phil phu phil phil
pht /pht 1 1 1 Xj Y, Z;
; P _"pht2 "pht2 J"phi2 "phr2 J"phr2 "hi2
2 da I, I, pht2 = 2 2 2 X, Y, Z,
P EL = Es: |, | ; i ;
pht2 Xpht Ypht 2 phi2 Ypht2 phi2 Ypht2 "phi? Ypht2
2 à 2 75 ^» ^ (15)
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 653