Hierarchical hangul character recognition with stochastic relationship modeling and candidate pruning = 확률적 관계 모델링과 후보제거 기법을 이용한 계층적 한글 문자 인식

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 274
  • Download : 0
Handwritten Hangul (Korean) character recognition is one of the most challenging problems in pattern recognition which endows computers with human cognitive capabilities. Since a Hangul character consists of several graphemes, the difficulty of Hangul character recognition can be compared to that of English word recognition, which is also known to be a difficult task. In Hangul, furthermore, the existence of many character classes of similar shape and touching between graphemes make the recognition more difficult. In particular, the touching between graphemes varies because Hangul graphemes are composed on a two-dimensional space, whereas Roman alphabets are composed in left-to-right order. These characteristics also make the recognition intractable. A great deal of computation is needed to discriminate the confusing character classes and to consider all possible grapheme combinations. In this thesis, two concepts, hierarchical relationship modeling and candidate pruning, are proposed to tackle those problems in handwritten Hangul character recognition. In structural character recognition, a character is usually viewed as a set of strokes and the spatial relationships between them. Therefore, strokes and their relationships should be properly modeled for effective character representation. For this purpose, we propose a modeling scheme by which strokes as well as relationships are represented by utilizing the hierarchical characteristics of target characters. A character is stochastically defined by a multivariate random variable over the components and its probability distribution is learned from a training data set. To overcome difficulties of the learning due to the curse of dimensionality, the probability distribution is approximated by a set of lower-order probability distributions by applying the idea of relationship decomposition recursively to components and subcomponents. Based on the hierarchical relationship representation, Hangul character recogniti...
Kim, Jin-Hyungresearcher김진형researcher
한국과학기술원 : 전산학전공,
Issue Date
181171/325007 / 000975001

학위논문(박사) - 한국과학기술원 : 전산학전공, 2003.2, [ x, 81 p. ]


Hierarchical character model; Stochastic character modeling; Handwritten Hangul character recognition; Candidate pruning; 후보 제거; 계층적 문자 모델; 확률적 문자 모델링; 필기 한글 문자 인식

Appears in Collection
Files in This Item
There are no files associated with this item.


  • mendeley


rss_1.0 rss_2.0 atom_1.0