International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV,
thesis named “A Mathematical Method of Communication”, in
which he presented a method to measure information based on
probability. In this thesis Shannon defined entropy, a concept
originated from classical stochastic physics to represent the
chaos of physical system, as the measurement of uncertainty
and it is calculated as follows:
Let X be a random variable with a set of possible
choices (A, A, … A, } with probability {P, ; P, 3 P, } :
then the entropy of X is
HXY= BR, PP Y=~3 PluP (D
;=l
I
When this approach of information measure is introduced into
quantitative map information measure, a natural thought is that
to give every kind of symbol on the map a probability, and then
the information content of the map can be calculated
corresponding to the probability. Sukhov (1967,1970) did the
initial work. He utilized a statistical model in which each kind
of symbol’s probability is calculated based on its frequency of
appearance in the map. His method is described as follows.
Let N be the total number of symbols on the map. M the
number of symbol types and F; the number of ith type, then
N = F + f de F, , the probability of each type of
symbol can be decided by
=
Il
= mn
~
Where P; is the probability of the ith type.
Then the entropy can be calculated through the probabilities
defined above.
M
H(X)=H(B,B,-,P,)=-> PlnP G)
i
i=]
This method firstly introduced information theory into map
information measurement, vct its fault is obvious. It completely
did not consider any position and topological information in the
map, which are very important components of spatial
information that we can get from a map. If symbols on the map
scatter in different manner (i.e. Fig.1), undoubtedly we can get
different amount of information from these two maps, yet this
statistical method mentioned above fails in this situation for it
can only obtain equal amount of information.
366
Part B2. Istanbul 2004
|
|
|
Figure 1. Two maps with same amount of symbols but different
distribution
For consideration of topological information more methods are
developed by researchers, some of them deserve being
mentioned here. Neumann (1994) proposed a method to
estimate the topological information of a map. In this method
vertices are classified according to their topological information,
such as how many neighbors they have. After the classification
entropy can be computed the same way by formula (2) and (3).
Yet in this method the classification of vertices is hard when a
map is relatively complex and the significance of the
classification is not so consistent with the real map.
Bjerke (1996) was not satisfied with such ways to define
topological information then he provided another definition of
topological information by considering the topological
arrangement of map symbols. He introduced some other
concepts, including positional entropy and metrical entropy.
"The metrical entropy of a map considers the variation of the
distance between map entities. The distance is measured
according to some metric’ (Bjorke 1996). He also suggests to
‘simply calculate the Euclidean distance between neighboring
map symbols and apply the distance differences rather than the
distance values themselves’. The positional entropy of a map
considers all the occurrences of the map entities as unique
events. In the special case that all the map events are equally
probable, the entropy is defined as H (X)=In (N), where N is the
number of entities.
Li and Huang (2002) proposed their consideration of
topological information on the map. In their paper they took
into consideration both the spaces occupied by map symbols
and the spatial distribution of these symbols. They divide
information about the features in the map into three types:
® (Geo) metric information related to position, size and
shape.
® Thematic information related to the types and importance
of features.
® Spatial relations between neighboring features implied by
distribution.
Based on this division they introduced Voronoi diagram to deal
with (Geo) metric information and spatial relations. A Voronoi
diagram is essentially a partition of the 2-D plane into N
polygonal regions, each of which is associated with a given
feature. The region associated with a feature is the locus of
points closer to that feature than to any other given feature as
shown in Fig. 3. As this region is determined by both the
feature’s size and the feature’s relative space, in some sense it
can represent the (geo) metric information and spatial relations.
Intern
After.
is tess
define
detern
The er
follow
H(M
As to
classif
and ca
All th
inform
the w
differe
differ
topolo
menti
some :
can re
or not
inform
the otl
they c.
the m
only r
map,
betwe
contai:
chang:
discern
all me
after t
metho
actuali
cannot
or is te
also c:
apply
amour