revision 
is study. 
some cul- 
ments is 
the ope- 
apping in 
iscussed. 
)y these 
1962. 
1, encom- 
° Agbor in 
is largely 
inor roads 
/ lOkm, is 
ities. The 
k forest. 
; than 50 
insur faced 
t of the 
’rocessing 
  
\rea 2 
  
3. IMAGE PROCESSING 
Contrast enhancement was done on the image be- 
fore it could be used at all. Based on the image 
histogram, contrast stretch was done by itera- 
tive use of the MAP function on the GOP 302 image 
processing system. A high pass convolutional fil- 
tering by subtraction of the laplace transform 
from the original image was later done to sharpen 
the high frequency details on the rectified images 
of the test areas. 
A part of the image was rectified using control 
points digitized on the Federal Surveys map sheets 
SW 299 and NW 299. The rectified area was approxi- 
mately 27km x 25km, from which the two test areas 
were later selected. The rectified images of the 
test areas are shown in fig. 1. 
Location of control points on the map and the 
image was a difficult task. Points, such as road 
intersections, which are normally used as GCP’s, 
could not be clearly identified on the image. 
On the other hand, most of the well defined road 
intersections on the image had not existed at the 
time of map compilation. 
In all, 19 points were identified, out of which 
13 were used for a second order polynomial trans- 
formation. The mean errors on 6 check points after 
transformation were 1.7 & 2.5 pixels in the row 
and column directions respectively. Resampling was 
done by bilinear interpolation, pixel size being 
kept constant at 10 meters. 
4. IMAGE INTERPRETATION 
The detectability of road networks and settle- 
ments in both test areas was investigated by com- 
paring the maps with the plots of the features. 
4.1 Test Area 1 
Figures 2 show the map of area 2 together with a 
plot of the features in the area. The upper part 
of the map was cut off so as to show exactly the 
same area on both map and rectified image. (See 
fig. 1(a)) 
4.1.1 Road network 
Using the cursor of the GOP 302, plotting of all 
identifiable roads was done on one of the graphic 
planes. Figure 2b shows the plot for Area 1. 
Main roads 
tous nma — 
According to specifications from the Federal Works 
Department regarding roads classification, main 
roads have a width of about 10 meters and are as- 
phalted. 
[Cr PE OR 
Hinlatad Compounds and Buildings 
Railways, Station, Siding, Level 
LOR er Te LTE il 
‚Kassa 
p 
Contous (VI. SOFt) ZZ 
PIA 
Contours [Ed 
Map Legend 
  
429 
The only main road passing through this area is 
fully identifiable and plottable. It has a width 
of approximately 10 meters and is asphalted. It is 
labelled AB in fig. 2b. 
v 
"BLIND RLLEY" 
ISOLATED SUILDINO 
  
Fig. 2(b): Plot of Road network in Area 1. 
Minor roads 
These roads are supposed to have a width of about 
8 meters and may or may not be surfaced. 
It was not possible to quantitatively relate the 
total length of minor roads identifiable on the 
image to the length on the map, as only a few of 
the roads shown on the image existed at the time 
of map compilation. 
For every road on the image that appears to end 
abruptly, it was verified on the ground whether it 
actually ends so or if further course of the road 
become unidentifiable due to some other reasons. 
It was found out that: 
. minor roads having width of at least 5 meters 
are generally identifiable. 
. along sections where road widths narrow down 
to less than 5 meters, the course becomes 
difficult or impossible to identify on the 
image. The decrease in width may be as a re- 
sult of part of the road being covered by 
weed or trees, as obtained at the points 
labelled " 2" in fig. 2(b), or it may sim- 
ply be that the road no longer continues 
with same width after a particular point. 
In the latter case, it appears to come to 
an abrupt end. 
Such points of apparent discontinuity are 
labelled "3" in fig. 2(b). 
. at some points, reflectance of the road and 
that of the adjoining buildings tend to 
"mix", thus making clear identification of 
the road impossible. These points are de- 
noted by "1" in fig. 2(b). 
It is estimated that up to 907 of the total length 
of minor roads having width of at least 5 meters 
are identifiable. 
Paths (Main and minor paths) 
Main paths are defined as tracks with width less 
than 4 meters. They may be occasionally motorable. 
A minor path, on the other hand, is an unmotorable 
track having width of less than 2 meters. 
Only a few paths could be identified on the image, 
typical width being about 4 meters. 
Identification was only possible where the paths 
pass through weedy areas. 
Paths in built-up areas were not identifiable.