For Z=0 condition these equations can be written in general
form:
P,X + PY +P,
9 Q)
= Ps X + PY + Pi
Ya Q
where 0 =P. X + PY +1, and P - (P, ... P.) — parameters
of perspective transformation.
In case of external orientation reference coordinates of test field
and their image coordinates are known. Then unknown
parameters P, , .., P; could be found by least mean square
solution from (2), which after multiplying by Q become a linear
function from P. So the vector of unknowns can be found by
one step.
The design matrix has the following form:
a 0 = 0 -X 0 Yxi XX
L —1 0 Ey 0 -X Yysi i
(3)
The estimation of vector of unknowns P is derived from
equation:
—
A'AP AIL, (4
T
where L, =(...~ x, c»
For determination of initial values for exterior orientation
parameters the condition of coefficient equality in (1) and (2) at
each matrix element a; is used.
After performing necessary calculation matrix A could be
written in the next form:
— Paw —Pyw
A Ls
f f
as cf IPS 0 (5)
y Jf
P,w Pgw 433
where w = —(X ja, + Yay + 2003)
„The values of unknowns w and a;3, a,3 as; are to be found. For
unknowns determination the condition of equality to 1 for sum
of squares for each row and each column of matrix A is used,
because A is rotation matrix. This condition allows to write the
next system of equations:
H PA hé |^ 0 0
w^ f 2 1
7 b = pip? 0i d Ü (6)
a 53 f^ 1
ass Pri o 9 1] i
0 1 1-1
from which w and a;;, 275 a5; are found.
Then angle parameters of exterior orientation are derived from
equations:
à — -arctg (a3; / A33)
Q — -arcsin (a53) (7)
K = arctg (a;;/ az)
The initial values for X» Yo Ze are found from system of
equations:
w — (X95; * Yod35 * Zod33);
Py — f(Xoajj + YoA12 + Z0913)/W;
resulting in:
x Pow! f
Y, |=a’| Bw! f (9)
Zo mM
Thus initial approximation for camera location Xy, Yo, Zo and
angle position a, c x are found.
3.2 Interior parameters estimation
The plane including 36 reference points is used as a test field,
the dimensions of the test field being about 1.5 m x 1.8 m. The
test field provides approximately equal targets density in
captured images. Reference points are marked with original
coded targets (Knyaz V., 1998) which provide robust and
precise reference point coordinate determination in the image
and reliable recognition of reference point number.
Using of coded contrast targets as reference points allows
achieving high degree of automation for interior parameters
estimation procedure. The only manual part of the procedure is
assessment of test field respectively camera in required
position.
2S0—
rm
In
ar
op