For an example shown in figure 6 that seven block
patterns can be extracted; two character blocks (“¡” and
“¤”) plus four digit blocks (“2”, “0”, “4”, and “9”) and a
single province-name block (“¦»Á¡¤®µ¦”). By this
example, members of set A are 25-feature value vector of
the character block of “¡”. Set B represents 25-feature
value vector of any character of reference patterns to be
measured similarity with. The feature vector of “¡” must
be compared with each character of reference pattern (“”
to “±” and “0” to “9”). The system identifies the input
character “¡” as identical as the reference character having
minimum distance to it. So, the distance of feature vector
of input character “¡” and which of reference character “¡”
should equal to minimum value (h(A,B) = min). Figure 7
shows the distance calculated for the input character and
each of reference character. Similar recognition procedure
is performed for a single block of province name where the
distances of input province-name block “¦»Á¡¤®µ¦” and
each of reference province-name (Totally 76 provincename
blocks but in this research, only 7 province-name
blocks could be provided) are calculated (See figure 8).