Such a function could be used in order to infer population error from sample error (from control points to compari -
son points).
9.2 Error Surface of the Photogrammetric Radial Error in Space
Let us arrange the radial errors in space as normals at different points on a plane. The upper ends of the normals
will then indicate an error surface. Using the regression function
+ - 8 +a X +a Y +a Z.
1 o 1 | 2.1 3 qj
fa YNXY"-3 X72 --3a YZ.
4 j i 5S 11 6 i1 1
2
2
Y +3 Z
1 19
2.
+a X., +a
7 1 1
8
we will find r as a function of the geodetic coordinates of comparison points, e I
Such functions could be calculated for the first model as n and for the second model as I: These functions could
be studied in many ways. Their derivatives will give the increase in error, and their intergrals error volume or
error mean,
Knowing.the proportions of standard deviations (Compare 6.13) it is possible to study the separate p z errors.
If r_ is a function of the geodetic coordinates X Y Z in the same system as n
transition from model one to model two. Let us look at the two functions n and I, as being polynoms or series.
, it is possible to study the -error
The difference (r, - n) or the quotient (,/T) will both indicate the error transition from the first to the second
e
model.
These error-transition functions could be studied in the same way by derivation, integration and subdivision of e 3
total error in p z errors etc.
The regression analysis resulted in the following functions where the base of the first módel is the unit of length.
1:3 500
+ 2
model 1 r=0 188 = 0 007+(Q 036% 0 013) Xj - (0 060* 0 027)X 5;
b
b
del 2 0. 1927 0.004 + (0.035? 0.006)X+ + (0. 1457 0.009)Y
model : y =0. - 0. + (0.035- 0.006 + (0. - 0.
b b
2 : 001 X 5.Yi «0.060X ^;
2-1 r =0.004 ~ 0. 1% + 0.14 Ji +0. 0X i
b b 2
b
1:6 000 Regression ended after step 1. No time for further regression
model 1
+ + + 2 +
model 2 r= (0 185- 0.008)#(0.258- 0.016)X, + (0.050- 0.014)Y; -(0.716* 9. 210)Y, Z.
2 He
b b b
2-1 -
Im
Al
I
Ink