Recognition And Application Of Handwritten Tibetan Character

China has a vast territory and a large population, which is composed of 56 ethnic groups, and Tibetan is one of the most important minorities of China. In addition to Chinese, Tibetan has become known for the longest history and the most literature of chinese cultural heritage, with it evolves over thousands of years. Nowadays, Chinese and English recognition technology has been quite mature and successfully applied in various fields. However, compare to Chinese and English, the recognition technology of Tibetan isn't such mature, research work and achievements on it is relatively little. This is due to some or more of the following as starting of research Tibetan Recognition technology is latter, the researchers need to be proficient in Tibetan language. The input of Tibetan stays in encoded using the keyboard. Obviously, this method is single, slow and cannot meet the needs of users. Compared with traditional input way of using keyboard, way of on-line handwritten input is more consistent with the human nature writing. With the further development of information technology, smart phones and other portable mobile computing devices are becoming more popular in recent years, on-line handwritten input method become an important human-computer interaction. It will have a wide social significance and market perspective.Based on the detailed introduction of the characteristics of Tibetan character, in this thesis, I do a lot detail research on handwriting Tibetan character recognition. The main contains are as follows:1. This thesis introduces background, current situation and the significance of Tibetan recognition in detail.2. This thesis comprehensively introduces the characteristics of Tibetan. Meanwhile, this thesis analyzes some of the unique features of handwritten Tibetan and the difficulties of the on-line handwritten Tibetan character recognition, and on-line handwritten Tibetan character recognition methods are further introduced.3. In this thesis I refined handwritten Tibetan character pre-processing steps. By this method, we can maximized keep the original information of handwritten Tibetan characterand filter out unnecessary interference and redundant information. That will prepare for feature extraction and recognition.4. This thesis introduces four kinds of features: directional element feature, strokes structure feature, gradient feature and Gabor feature. Comparing them by experiments, get two better one. Meanwhile, in order to use both of on-line recognition method and off-line recognition method, two fusion methods that prove the recognition rate has been improved by experiments are proposed.5. On the Android platform, we develop handwritten Tibetan character software and expand it to be an input software.