238 
PHOTOGRAMMETRIC ENGINEERING 
h(n)max — h(n)min = r, the range. 
With rectified data, three x', y' coordinates 
predict the selected upper limit, lower limit, 
and the center of the resultant search range 
(r) in the conjugate photograph. If a corre 
sponding matched point is located, where 
f(AD)' is the first absolute maximum along 
this range, it is accepted as a matched point 
and tagged a “measured” point; if not, the 
predicted elevation h Xtv is accepted as the 
correct elevation. In either case, the derived 
elevation is tagged to indicate whether it is a 
predicted elevation. The predictive process is 
then repeated in its entirety with the next 
NXM area in the rectified'“x” direction. A 
further opening of search limits (r) then takes 
place when the predicted parallax is above 
the approximate mid-elevation level in the 
model. In steep terrain, whenever the last 
“measured” point in the x direction, A(l), of 
Figure 12 is a predicted value, the range opens 
to a larger selected value. 
The next program section is for contouring 
the elevations determined by the predictive 
process. This routine initially corrects each 
predicted point for relief displacement. These 
ortho-corrected elevation points are dis 
tributed over the entire photograph at about 
0.090 inch for an NXM area of 36X36 spots. 
The program then interpolates between these 
corrected matched points to produce an 
elevation for each corner point (nearest the 
nadir point) of a 6X 6 spot area. The elevation 
values are then given their appropriate con 
tour designations which are examined to 
determine where the contour lines should be 
placed. A contour line, appropriate in width 
to the five sizes used, is placed in the 6X6 
spot area that contains a contour interval 
change. (Other stored symbols can be en 
tered at this point.) The formed contour lines 
can then be transferred to magnetic tape and 
printed by the scanner-printer as a contour 
manuscript. 
Direct Orthophoto Program 
When the x, y tabulations of point eleva 
tions are made for relief distortion shifts, the 
amount of displacement involved is recorded 
as an amount in x and y shift. Both the un 
ortho-corrected x, y point coordinates and the 
ortho-corrected x", y" coordinates are stored 
in table format. The vector distance between 
these two points represents the ortho-dis 
placement. The displacements are then con 
verted to a shift list which specifies the dis 
placement required to move each picture ele 
ment in the prime photograph to its new 
ortho-corrected position. The list-generating 
routines indicate where in the photograph a 
need exists to extract or insert photo points 
(i.e., photo detail) and where photo data is to 
be re-positioned. (When ortho-correcting un 
rectified photographs, corrections for the tilt 
and scale distortion determined from resec 
tion and orientation would be combined with 
the relief displacement due to ortho-correction 
to produce an orthophoto map.) 
In using this shift list, the photographic 
data are now processed one scan at a time, 
and the necessary set of spot corrections in x 
and then in y are made. The ortho-correction 
program computes the x and y vector shifts of 
the radial relief displacement vector of a 
point (Ar = fxh/p-h) at rectified photo-scale 
and then corrects for datum-scale, “p” is the 
principal distance of the point from the nadir 
(more exactly the isocenter), and “h” is the 
elevation of photo scale. By compressing or 
expanding the data a point at a time, the best 
quality of an orthophoto is obtainable. These 
data are then stored on tape for printout of 
an orthophoto. 
(See Summary on the following page.) 
Conclusions 
The Digital Automatic Map Compilation 
System has produced maps and orthophotos 
of sufficient accuracy (C-Factor greater than 
250) by one system and sufficient speed per 
model (less than 60 minutes) by a second 
system unquestionably to prove technical 
feasibility for producing maps and ortho 
photos by digital techniques. 
It is apparent that the precision, resolution, 
and compilation speed of the present photo 
digitizer and computer system can be im 
proved through refinements and redesign 
which are within the state of the art. 
The anticipated goals are a C-Factor of 
1500 and 1 to 2 hours per model of data proc 
essing time. Based on extensive experimenta 
tion and planning, it is concluded that these 
goals are realistic. 
A digital mapping system that is economi 
cally justifiable, as well as technically feasible, 
can be developed within the next several 
years, as an extension of this project. 
Acknowledgments 
Many individuals contributed to the suc 
cess of this project over the past five years. It 
is a pleasure to give acknowledgment to A. L.