Discrimination of Ambiguous Hangul Character Patterns Using Fuzzy Membership Function

초록

In this paper we consider a combined character as a meta symbol which may be recognized ambiguously as a multi-category. The patterns that may be recognized as a multi-cayegory are analyzed via the fuzzy membership functions. In analysis, the used attributes are as follows: a slope and a length of a primitive, and relative geometric positions among the primitives. Where primitives are extracted from each character pattern. A set of different characters that have the ambiguity is treated as an ambiguity category set. Elements, i.e., character patterns, belong to an ambiguity category set have commonly the subpatterns that result in the ambiguity. The subpatterns are decomposed into basic primitives, and their attributes are parameterized to the defined fuzzy membership functions. The obtained values by applying the fuzzy membership function are used to observe wether the ambiguous pattern belongs to a certain category or not.