{ :- 
Message Nature FAR Encoding process chart word |Length 
(0-60) e 
[p 
$, 000 0.40 1 | I 
32 ool 0.18 00] 3 
9s oto 0-10 oil 3 
$9, oll 0-10 0000 4 
js 100 0.07 0100 4 
9% iol 0.06 0101 4 
97 Io 0.05 90010 S 
98 H1 0:04 00011 5 
  
  
  
  
  
  
Fig. 4 An example of Huffman encoding 
small, 
(2) Assign the digits O and 1 to the two mersages 
with the smallest probabilities respectively, then 
calculate the probability summation of that two mes- 
sages. 
(3) Take the above-mentioned sum of probabilities as 
the probability of a new message, put it in the rest 
original messages and rearrange the reduced set in 
accordance with a sequence of declining probabili- 
ties. 
(4) Repeat the previous steps (2) and (3) till the 
summation of two probabilities is equal to unit at 
last. 
(5) Set out the latest assignment of digits, trace 
the way adverse to the encoding progresa of indivi- 
dual messages and pick the allocated binary digits 
at all steps in eucceesion. Finally the digital 
etrings of O and 1 placed in such a way are taken 
as the code words corresponding to the individual 
messages respectively. Figure 4 illustrates an ex- 
amnle of Huffman coding. 
The principles of optimum coding which digital 
image has to observe are: (1) the uniqueness of 
decoding, and that any short code word does not 
form a prefix of the longer one; (2) the mersage 
with larger probability deserves a shorter code 
word, and the message with smaller probability a 
larger code word. Huffman coding meets the above- 
mentioned elements at all. It is preferable to uti- 
lize run length coding in cooperation with Huffman 
coding. Generally run length coding ies conducted 
rrior to the implementation of Huffman coding so 
that one can reach the coding efficiency as high 
as possible, 
434 
5. THE COMMUNICATION MODEL OF DIGITAL 
MAPPING 
In Figure 5 there is showed a model of communica- 
tion system that is discussed in Information theory 
. It ie also appropriate for dealing with the other 
information transmiesion systems, so the diagram in 
Figure 5 is a general information processing model. 
We use the common flow scheme of digital mapping as 
an example to explain the actual meaning in each 
block of that diagram. 
In the information source of digital mapping there 
are unrectified airphotos, specification parameters 
of camera, coordinates of ground controls and other 
initial data. Aerial imagery and geodetic coordina- 
tes belong to the different record types of geogra- 
rhic information. The image in airphotos is merely 
some approximate impression of the real rcenery ap- 
rearance of the earth surface. The original airpho- 
tos often involve various deformations due to both 
geometric and optical errors in the aerial photo- 
grarhic process. In those errors there are not only 
the systematic but also the accidental ones. This 
shows the information source consists of signals 
and noises from the beginning. One can not adopt a 
simple communication mode before data preprocessing. 
Information destination is the output results of 
digital mapping. They usually include the topogra- 
phic and thematic maps, the orthophotomaps and the 
other documentation of iconic represetation which 
are rendered to urers. These products contain the 
reliable records of the geographic information re- 
lated to the aerial survey area. They may become a 
significant input part of the regional geographic 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996 
ei, 
me! 
ho 
nic 
tei 
be: 
Dec 
ces 
is 
phe 
to 
lir 
of 
etc 
the 
the 
The 
rer 
con