48
Table 1.
Tube S19930
Tube S19931
Standard Deviation of the Nominal
Versus Calibrated Grid
Similarity
Affine
X
=
4.9 pm
X
=
3.7 ym
y
=
3.6
y
=
1.6
R
=
6.1
R
=
4.0
X
=
4.0
X
=
3.2
y
=
2.6
y
=
1.3
R
=
4.7
R
=
3.5
Values are given at tube scale
1 ym at vidicon tube = 3.9 meters at ground
micrometers at tube scale (1 micrometer equals approxi
mately 3.9 meters at ground scale). It is apparent that
the affine transformation reduces the residuals 2 to 4
micrometers (8 to 16 meters at ground) below those of the
similarity transformation. Also the y direction has
considerably less variability than the x direction. When
comparing these results with those of Table 1 it appears
that the result of producing the image film by the Electron
Beam Recorder adds only 1 or 2 micrometers (4 to 8 meters
at ground) to the residuals.
Nominal Positions Versus Measured Positions. When the
RBV image is re-created from magnetic tape it is
"corrected" to the calibrated positions of the reseau
intersections. Therefore, it might be expected that an
adjustment of the measured intersections on the film to
the nominal values would yield a larger variability than
when adjusted to the calibrated intersections. The results
(at tube scale) are shown in Table 3.
Comparing these results with those from the adjustments of
measured to calibrated, one sees that the similarity
transformation produces slightly lower residuals for the
former but the affine produces larger residuals. It
appears from this if one confines oneself to a similarity
transformation operation (e.g., simple enlargement) the
nominal intersections will be sufficiently accurate.
Image-to-Ground Adjustments
Map Control. Ground control was taken from 1:24,000-
scale topographic maps. The precise locations were chosen
by comparing well defined map points with their corre
sponding image points on the RBV film as viewed with a
monocomparator. The Universal Transverse Mercator (UTM)
coordinates were read and recorded to five meters using a
metric coordinate reader.
For 1:24,000-scale maps that meet National Map Standards,
one may expect errors in position as large as 13 meters,
which along with errors of identification approaching 10
meters can cause errors in control as large as 25 meters
and an RMS of 12 meters.