8
the line between the two camera stations. A high
height estimate is indicated, where the image in
the center of the held of view on the second dia
positive is shown as P 0 (instead of the correct
position, P). The image of the estimated point,
P e , on diapositive 1 is displaced G z dZ from the
image of P, and by GudZ from the image of P n .
In practice it is expedient to make the correction
using G H dZ at the first diapositive. The ortho
photo signal is taken from the second diapositive ;
the displacement of P 0 from P is ordinarily so
small as to produce negligible error on the ortho
photo. By assuming AZ = 1 (i.e., equal to the
increment of the height counter), the required
height corrections are readily found. Thus,
dx - G x dX + GydY + G z
and (5)
dy = H x dX + HydY + H z
In Eq. (5), dX and dY are found from the cor
responding equations for the second diapositive
by taking dx' = dy' = 0 ; that is,
0 = G'xdX + G'ydY + G'z
and (6)
0 = H'xdX + H'ydY + H'z
When Eq. (6) is solved for dX and dY and the
results substituted into Eq. (5), the differential
shifts, dx and dy, corresponding to a unit altitude
error are determined. These values are outputted
to the analog equipment. Because only a very
modest accuracy is required, simple approxima
tions are possible and only infrequent updating is
required. The calculation represents a trivial part
of the total computer problem.
The computer also outputs a signal to control
the intensity of the scanner exposing the altitude
chart. For this purpose, the altitude is divided by
three times the desired contour interval and the
remainder, R, examined for assignment of the
printout level. The level is assigned as follows:
0 < R < 1/3 scanner off
1/3 < R < 2/3 scanner medium
2/3 < R scanner bright
This creates a chart (shown on the first page)
showing well defined altitude bands. The rotary
sequence of densities permits an unambiguous
assignment of altitudes, starting from a point of
known altitude.
REDUCTION OF V PARALLAX
Creating an adequate stereo model for compila
tion requires that both X and Y parallax be
removed at a number of points in the stereo field
so that coordinates of common points in the field
can be accurately ascertained. The height-error
sensor removes X displacements caused by height
errors; therefore, it is directly applicable to the
automatic removal of X parallax in the compara
tor mode. The removal of Y parallax displace
ments requires additional circuitry.
Sensing of Y errors is achieved, in effect, by
introducing an artificial Y parallax analogously
to the X parallax provided by the delay lines in
the height-error sensor (Figure 6). The Y paral
lax is introduced on a time-sequential basis by
shifting the scan in Y on one diapositive, posi
tively for a defined time and then negatively for
an equal time. The output of the associated corre
lator is switched into a positive input of an inte
grator during the positive shift, and to a negative
input during the negative shift. If there is no Y
parallax at the point, the inputs during the two
time periods are equal, and the integrated output
is zero. If there is any Y parallax, the correlator
will have a larger output when the artificial par
allax partially cancels the image displacement and
a lower output for the other half cycle where the
two components are additive; therefore, the
integrated output provides a measure of the Y
parallax.
In the UAMCE, a positive and negative thresh
old detector (similar to that used with the height-
error sensor) provides digital error signals to the
computer for reduction of Y parallax during com
parator operations. For automatic compilation
operations, the integrated output is used to deflect
the center of one of the diapositive scans in a
direction to reduce Y parallax; this ensures high
sensitivity in the height-sensing circuitry, even
with a poor model.
SCANNING PATTERN
It has been noted that the principal scanning
for automatic height determination should be
along a line parallel to the line joining the two
camera stations (i.e., on an X-directed line). This
must be augmented by a slow scan perpendicular
to the fast scan (on a Y-directed line) to produce
an effective area scan providing a TV-like cover
age of each small area. This ensures continuity of
the image for both the operator and for automatic
tracking. The signal from one scanner is used to
recreate the scene element for exposure of the
orthophoto film sheet.
For automatic operations the time required to
sample a given area must be kept as low as pos
sible. For this reason the Y scan is made in an
unusual pattern. Every line on the positive side
of the center of the scan is followed by a corre
spondingly positioned line on the negative side.
The scanned area is covered very coarsely at first,
then gradually filled in. A total of 128 lines is