Preprocessing And Feature Extraction Of On-line Uyghur Handwriting Recognition

Processing and recognition of letter information with using computer have been an active research topic in the fields of pattern recognition. Along with the rapid development of mobile digit devices, it has been difficult to satisfy people's demand by using the traditional keyboard to input characters to the computer. Also along with the advancing of people's living standard, computer is playing a very important role in everyday life and work. Uyghur letters account for quite large proportion in the process of information in Xin Jiang Uyghur Autonomous Region, unfortunately we could not be able to find research on On-Line Uyghur Handwriting Recognition from articles in existence. Research and relative work should be offered in order to accelerate extension of information as well as create convenience for people.Main content of this article is preprocessing and feature extraction of On-Line Uyghur handwriting. According to characteristics of Uyghur handwriting letters bring forward relative techniques of preprocessing, feature extraction, vector quantization with Excellencies as well as disadvantages. Presently to reach the goal of recognition of handwriting letter which has to pass through five stage in most of HMM based handwriting recognition system, these are preprocessing, feature extraction, vector quantization, training model and recognition. Among them we only demonstrate our work on the previous three stages, and mention some technical aspect and some idea of last two stages. During preprocessing, feature extraction and vector quantization stages, according to the characteristics of Uyghur handwriting letters, we made careful study and analysis, implemented corresponding algorithm with suing Visual C++6.0 program environment.