Single-character segmentation for handprinted Thai word

In this paper, a new scheme for extracting handprinted Thai characters from a given word script is proposed. The principal domain of application is the character-type identification (vowel, consonant and tonal symbol). The type of character plays an important role in many handprinted Thai character recognition systems. The approach is based on global feature analysis, for instance, consonant character segmentation using a projection profile. Firstly, a brief description of the Thai character and Thai word structure is described. Thai words are consequently extracted from the input document. The slant angle for each word is computed and normalized by using an affine transform. In this word, the characters are extracted and character-types are identified. Eventually, an isolated single-character and its type is obtained. In the experimental results, 100 samples of documents (handprinted Thai words) written by 10 subjects are implemented