Online Handwriting Thai Character Recognition

This research is a study of structure and method in handwriting Thai characters for conversion and computer use. This paper covers the creation of an online handwriting recognition system through the use of linear interpolation approach. The character sampling is done with 10 significant points for the vector creations of angles and rotation directions. The linear-interpolated characters will then be compared with an XML, database in which contains predefined character information. The best matched character is then further analyzed for character uniqueness such as tail and size before concluding the found character(s). This system supports the writing of different character styles including slanted, scaled and rotated characters. The average recognition rates are of 90.88% correctness for matched character and 98.75% correctness for a candidate group of found characters in which is satisfactory and effective for Thai handwriting character recognition.