tion by the desired contour interval. The operating
point moves up (or down) the terrain slope to the
next contour, as indicated in Figure 9. When indexing
at a boundary, the system moves (approximately)
along the boundary to the next contour line. When
indexing on contour closure, the system moves di-
rectly up the slope from the closure point to the next
contour. Indexing in profiling involves moving to a
parallel profiling line at a specified distance from the
profiling line just completed.
U-- ---4
Model or Photo
Boundary
Indexing at Boundary
Start
Indexing on Contour Closure
Figure 9 Automatic Indexing During Contouring
Automatic indexing between parallel profiling lines
can occur in this way until the entire area has been
profiled. Automatic indexing between contours can
occur in this simple way until the top of a hill (or
bottom of a valley) is reached. Contouring is nor-
mally started at the lowest (or highest) contour eleva-
tion in the model. At each contour elevation, only
the connected part of the contour line will be plotted
automatically; disconnected sections must be plotted
separately. When a saddle point on a hill is reached,
the system will automatically continue contouring up
one peak and ignore the others. Upon reaching the
top of each peak in the stereomodel, the operator
must (in present systems) stop automatic plotting,
locate the lowest unplotted contour around another
peak, and restart automatic plotting.
Automatic plotting can proceed along successive
lines as described above as long as terrain image con-
ditions are good. If image conditions become poor,
the terrain surface may be lost by the stereopercep-
tion system. When the terrain surface is lost, auto-
matic plotting must be stopped to avoid plotting of
grossly incorrect data. Detection of lost situations
during automatic plotting is performed primarily on
the basis of low measured correlation. Lost condi-
tions occur most often in areas which lack terrain
detail, such as lakes and large featureless fields. Lost
conditions may also occur when the terrain slope
changes suddenly, such as in crossing the top of a
sharp ridge, particularly if there is a large shadow area
associated with the ridge.
The computer can be programmed to attempt
automatic recovery from lost conditions by perform-
ing a systematic search. A three-step automatic search
strategy, which has proved to be highly effective in
the automated analytical stereoplotters, is: (1) lift the
stylus and stop the plotting to allow parallax and
slope measurement errors to be corrected, (2) make a
series of small trial motions based on possible changes
in plotting line direction, and (3) move forward in the
original direction (at constant elevation for profiling)
for a larger distance. The first or second steps are
generally quite successful in recovering from lost
conditions associated with sharp ridges or valleys
without generating a large gap. The third step is most
useful for bridging rivers and lakes in the profiling
mode. In some situations, of course, the search
strategy fails, and the computer must then signal the
operator for manual assistance.
AUTOMATIC PLOTTING PERFORMANCE
A typical contour manuscript, automatically plot-
ted by an automated analytical stereoplotter, is
shown in Figure 10. This chart was plotted from
vertical photography at a scale of approximately
1:48,000. The flying height was 24,000 feet. Within
the area shown, three types of terrain geometry
occur: rolling hills in the upper portion, a flat region
at the central right, and a more sharply mountainous
region at the lower left.
In the rolling hills, the system plots rapidly—at an
estimated three to five times the speed of a human
operator. The occurrence of transient errors and lost
conditions is negligible, and the automatically com-
piled manuscript requires almost no editing.
In the flat areas, the system plots less rapidly and
generates more gaps, due both to poorer image detail
in the open fields and to transient elevation errors
which accompany the parallax-integral steering.
CHAPELLE, WHITESIDE, AND BYBEE
veux agi,
PA FANGS
Iu
are
tic
tir
tir
tic
th
th:
ga
tir
tu:
th:
th
er:
rai
ste
sp
th
tic
the
Cc
ele
ac
INA
gr: