236
PHOTOGRAM MET RIC ENGINEERING
The general procedure of matching will be
described in geometric terms. The left photo
is considered to be divided into square areas,
and each square, in turn, is considered in an
attempt to find a matching area in the right
photo. All of the squares forming a horizontal
(parallel to .v-axis) band are considered before
/Considering squares in another band. The
second and third stages of matching are simi
lar in operation to the first stage.
There is enough room in the computer
memory to store the parallax approximations
for parts (B) and (C). The parallax approxi
mations found by part (D), with interpola
tion where necessary, are written on a tape as
each band or strip of the left photo is matched.
Each parallax difference on the output tape
corresponds to a 6X6 spot area of the full-
resolution left photo.
Ortho-correction and Contour Program
An orthophoto map with contours is pro
duced after the correlation process. It repre
sents an orthogonal view of the ground in
which each object appears as though it is
viewed from vertically above. Image dis
placement caused by relief is one of the more
significant errors in a real photograph result
ing from the perspective view of camera-lens
combination. The orthocorrection program
corrects for image displacement resulting
from relief and, in conjunction with the
W ILD reproducer unit, generates an ortho
photo map and/or a contour overlay.
The 6X6 photo spot increment (15-mil
square) on the left rectified photo tape is
moved to its orthographic location on the
basis of its elevation relative to the nadir-
point. (»roups of six scans of rectified data are
transferred into an input buffer where the
elevation of each 6X6 spot area is computed
from its associated parallax, previously de
rived from the correlation phase and stored
on a parallax tape. Each 6X6 spot area is
assigned an elevation which, in conjunction
with a series of algorithms, places a contour
marker in each area whenever the elevation
moves to the next higher or lower contour
level. The contour placement algorithm
verifies that each contour line is continuous
and closes upon itself. After a correction fac
tor is applied to the coordinates of each ele
mental area, the data are shifted from the
input buffer to its new' or ortho-corrected
location until it is written on tape. The format
of the output tape contains the ortho-cor
rected photographic data in the low-order
Maximum of density
difference function is
assigned to point P
(x,y).
N x M area in which point to be
y “* M correlated lies, P x,y.
h(l)*
h(0).^
/1
N
i
h(2)
h(3)
•
h(4)
\
Known elevation h(2) of the second of the
four previously correlated points used to
predict h (0).
Fig. 12. Predicting parallax of P(x, y).
bits of a character, with the contour spot
markers appearing in the high-order bit.
The present format for recording contour
information on an overlay is as follows:
1. Every fifth contour interval receives a
darker shade of gray that makes it more
discernible than the other contours.
2. Each contour line is represented by a
unique symbol to differentiate it from
the adjoining contours. Five different
symbols are available. A contour line is
generated by interconnecting a series of
6X6 spot areas in which the photo
graphic detail has been replaced by an
appropriate symbol.
These marks are placed at each point where
the parallax difference changes to a new con
tour interval, producing a contour map in
digital form.
Tic Mark Program
Planimetric maps usually contain identify
ing information that shows the position of the
photographs relative to some ground co
ordinate system. The Tic Mark program
inserts markers at 1,000-meter latitude and
longitude intersections on the orthophoto or
contour map. Based upon ground-control
information and flight angle, an imaginary
grid system is constructed to the scale of the
map and oriented in a north-south direction.
Those photo locations that coincide with a
grid intersection are marked so that they can
be identified on the final contour map.
DAMC PREDICTIVE SYSTEM
The major operational parts of the predic
tive system are essentially completed (Figure
10). The scope of this effort was to develop
and show feasibility for a second experi
mental programming system, and thereby to
increase the system flexibility by providing