Road 
cted in 
for the 
y. The 
the St. 
ear sea 
reaches 
en in a 
re then 
gitized. 
major 
power 
1gs and 
mode. 
evision 
. The 
] as the 
gitized 
rdinate 
tions at 
dinates 
ial and 
ions at 
es was 
curacy 
0 meet 
acy of 
e North 
tem for 
Ires 
bly 
fall 
tric 
or 
ements 
t mean 
m. 
e Land 
curacy 
ip Base 
indards 
res” 
£ 
nal 
tive 
req) 
tion 
real 
VISE in 
curacy 
digital 
ground. 
of V2 to 
is equal 
RMSE 
E for a 
e5.0m 
scale. 
3.3 Test Results 
The results of the assessment are shown in Tables 1 to 3 for 
the 1: 50 000 scale NTS map and in Tables 4 to 6 for the 
1:10 000 LRIS map. The type of features mapped, the 
number of points tested along each feature (No. pts.), the 
root mean square error in metres (RMSE) and the percent of 
test points where the coordinate differences exceeded a 
certain tolerance value, are listed. 
4 ANALYSES OF RESULTS, CONCLUSIONS 
AND RECOMMENDATIONS 
Three schemes were developed and implemented for 
detecting changes in the planimetric content of digital maps 
and for revising the changes. All operations are performed 
in a GIS environment, which has been equipped with 
vector/raster handling capability. This arrangement has a 
number of advantages: 
All operations are performed in a commercially available 
GIS workstation. No additional, special hardware is 
required. 
* The revision process is an extension to the existing GIS 
functions, so that an operator can perform these tasks 
with no or very little additional training. 
* The pliotogrammetric operations, which are incorporated 
in the processes, are largely transparent to the user and no 
knowledge of photogrammetry is required on the part of 
the operator. 
* All measurements are monoscopic so that the ability of 
stereo vision is not a prerequisite. 
The results obtained for the revision test of the 1: 50 000 
scale map are entirely satisfactory for all three schemes. A 
total of 1476 points were tested and with the exception of 
four along the forest boundary, none exceeded the 25 m 
tolerance prescribed in the specifications. Forest boundary 
is not considered as well defined feature. This result was to 
be expected because the photo scale was larger than the map 
scale which left a larger margin for identification, measuring 
and instrumental errors. It is not uncommon to take 
photographs at such a large scale for 1: 50 000 scale 
mapping. The operating ceiling of most aircrafts used for 
aerial photography is 25,000 feet (7,625 m) and the photo 
scale obtained from this altitude with a wide angle camera is 
1: 50 000 [Slama (ed.), 1980]. 
The results obtained for the 1: 10 000 scale map are less 
favourable. Out of the 1701 points tests 31%, 6% and 7% 
exceeded the 3.3 m limit and 1096, 0.496 and 396 exceeded 
the 5.0 m limit set for the 9096 error using the piecewise 
rectification, the DEM corrected tracing and orthoimage 
tracing schemes respectively. The following explanations 
are offered: 
* The pixel size of the digitized photograph is 3.0 m which 
is the same magnitude as the tolerance set for the map 
accuracy. 
* The map accuracy standards set by NBGIC are 
considerably more stringent than the customary 0.5 mm 
tolerance at publication scale which would correspond to 
5.0 m on a 1:10 000 scale map. With the exception of a 
few features in the piecewise rectification case, this 
standard was satisfied. 
+ Difficulties were encountered in some section of the image 
to find well defined features around the area to be revised 
for the rectification. 
* This project was the first experiment for testing the three 
map revision schemes on real data. It was performed by a 
novice operator without the opportunity of gaining the 
necessary experience needed to generate consistent 
results. 
835 
Nevertheless, this experiment is a good preliminary indicator 
of the practical value of the proposed map revision schemes 
and points to certain modification to be made in the future. 
In particular the following conclusions and recommendations 
were reached: 
All three schemes are feasible alternatives for the revision of 
digital maps. The orthoimage tracing provides the highest 
accuracy, followed by the DEM corrected tracing and the 
piecewise rectification. Data requirement (DEM) and 
preprocessing time are in direct relationship with the level of 
accuracy obtaineable and the cost incurred. 
The piecewise rectification scheme, which is the simplest 
and most economical of the three is perfectly satisfactory for 
the revision of medium scale maps. For large scale maps the 
DEM correction or the orthoimage scheme are 
recommended. It is expected, however, that as experienceis 
gained, the piecewise rectification results will improve. 
The piecewise rectification is a satisfactory method for the 
updating of resource inventory maps at all scale. Here the 
map accuracy standards are less stringent than in basic 
mapping. 
It is recommended that for the revision of large scale maps 
the photo to map scale ratio be reduced to about 2.5:1 and 
the resolution of the digitization be increased to about 500 
dots per inch or 0.05 mm. This resolution can still be 
achieved with medium priced scanners. The performance of 
scanners is steadily improving and the price is becoming 
more affordable. 
More experimentation is needed to evaluate the full potential 
of the map revision schemes presented here, and to refine the 
procedures used. There are definite plans to do this. 
ACKNOWLEDGEMENT 
This research and development work is being funded under 
the Canada/New Brunswick Subsidiary Agreement on 
Industrial Innovation and Technology Development and by a 
National Sciences and Engineering Research Council, 
Canada grant in aid of research. The authors also wish to 
thank Universal Systems Limited for their collaboration. 
REFERENCES 
Derenyi, E. and R. Pollock (1991). “Design and 
Development of a Heterogeneous GIS.” CISM Journal 
ACGC, Vol. 45, No. 4, pp. 561-567. 
Masry, S.E. and R.A. McLaren (1979). “Digital Map 
Revision.” Photogrammetric Engineering and Remote 
Sensing, Vol. 45, No. 2, pp. 193-200. 
NBGIC (1991). “Province of New Brunswick Land and 
Water Information Standards.” New Brunswick Geographic 
Information Corporation, Fredericton, N.B., Canada. 
Slama, C.C. (ed.) (1980). Manual of Photogrammetry. 4th 
Ed. American Society of Photogrammetry, Falls Church, 
VA, U.S.A.