| Summary: | 碩士 === 國立屏東商業技術學院 === 資訊管理系 === 95 === The technical development of Chinese handwritten recognition has reached a bottleneck. The reasons of affecting the recognition performance can be the user’s stroke writing sequence and cursive writing. To improve the recognition rate effectively, we adopt the Chinese phonetic input method, which is naturally used by the native speakers and uses the hidden strokes as the extra recognition information. In this thesis, based on the phonetic symbol recognition, two application systems, the on-line handwritten EPS(effective phonetic sequence) recognition system and the on-line Liang-Mian-A Chinese recognition system, have been proposed. Note that the on-line Liang-Mian-A Chinese recognition system is used to improve the problem of paronym. According to the experiments, the recognition rate of on-line handwritten phonetic symbols is 98.05%, and those of on-line handwritten EPS and on-line handwritten Liang-Mian-A codes are 97.34% and 100% respectively.
|
|---|
