^9
ne
P SE Era TR I om T. MÀ ;
pe LEAST QUT NEN Da E p eo EROS ERA TN 1
17.
in Table 3. This is similar to the parameter used by Thompson. The
graph however, revealed no trend between 1/0 and systematic errors
and has thus not presented on this paper. In fact a scanning of Table
3 will reveal the variable nature of these errors. There are many
unknown variables inherent in the results of Table 3. e.g. variation
in image quality with image position, precision of stereocomparator -
etc., and hence such comparisons are unreliable. The major trend in
Table 3 is larger errors with higher contrast.
3.3 Conclusions
Based on the work discussed, the following conclusions are made
(1) Pointing to edges of objects results in a subjective
location of the edge which is usually towards the less dense side of
the edge
(ii) The errors of both Thompson and Welch and Halliday appear
to be in general agreement, and are larger than 100 secs of arc
depending on target contrast and image quality. It is not known
however whether these errors are purely a function of optical magni-
fication. If they are, errors in excess of lOym (i.e. 4 times those
derived by Welch and Halliday) will occur if the optical magnification
is 10X.
(iii) Errors will be substantially larger for high contrast than
for medium contrast. Medium contrast(approximately 2.5:1 to 5:1) will
result in the optimum image characteristics for edge pointing.
(iv) Considering the unreliable nature of edge pointing it is
recommended that some '"ground truth" data be made available to the
operator so that his personal bias can be determined.
4. Interpretability of Photogrammetric Details
An observer's ability to correctly interpret details on photo-
graphic images is the task on which a minimum amount of research has
been carried out. A study of this task requires the consideration of
substantially more variables than is the case with purely measurement
tasks such as target and edge pointing. Indeed interpretation has as
a pre-requisite a period of learning on the part of an observer. This
process of learning is related to the type of objects to be interpreted,
and cannot be considered in this paper. Aspects of interpretation that
will be considered, are only those of image quality which affect the
recognition of simple objects.
Hempenius (1968) in a most comprehensive paper discussed many of
the aspects affecting interpretation of details from photographs, and
it is not necessary to cover these aspects in this paper. Many of the
aspects referred to by Hempenius have not yet been investigated,
especially the psychological factors, and researchers interested in
these aspects are referred to this paper.
—————————— MEN
DEED UR SR PET TEASER ASS Acta Un OL A Eg