Jen-Jer Jaw
je more With the above system formulation, the least squares method would lead to the solutions as follows
nstraint
'chastic The estimation of the parameters
fication
surface A = Z 4 > a y
ollows, x =1N Nu MaWNn -NaNu Mo) NaNy LG UN Mal Na Na Ni Ma) EC, (7)
A zi x z
(5) x, -[-(N5, - NN, Nyy) 'NoN 1 '1-C + IN, — Noy Ny EN) *1-C, (8)
T
Ny =A RA, C, =A, Py,
where Ny, =A PA, =Ny
T — —
Ny = An PA +An PoAn, C4 S Ag! Py, t Ax. Bow
Jetailed
The dispersion of the parameters,
2 ^ = - = :
S, DE Na Nu. No» - NaNua Nip) ™ Nyy Ny (9)
Sei. Df = (Vas Na Wi)” (10)
' system
The predicted residuals:
(6) : ^ ^
€ 7 Y - A4 347 Ai; X, (11)
e; -—P, B^ P. (As X -7W)s xp, * Q,,)B" P fion X — W) (12)
The estimated variance component:
T. T.
»
vtae esq Ao Be
2k - m —6p — 3n
o
^
The estimated dispersion of the exterior orientation parameters ( X, ):
" bs | 29/97. p | (14)
^
The estimated dispersion of the object point coordinates (X, ):
bx f= Wem Di b (15)
32 Best Fitting of Object Points and Surface Points
The system solved by the above least squares method not only estimates the unknowns of object points measured via
photos but also adjusts the registered surface points combining collinearity property and surface plane constraints. Thus
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 447