Skip to Main Content
Probing into the features of Manchu language especially its characters or letters, the stroke elements of primitive Manchu are defined. Then, a system of offline handwritten primitive Manchu character recognition method based on strokes is built. Briefly, a pre-process for primitive Manchu is operated to get single word with digital image process methods, which then decompounds the word to stroke elements, the stroke elements sequence is obtained by using statistical pattern recognition method and the Manchu-Roman code is obtained by using fuzzy string matching algorithm. Simulation results showed that the result is effective based on the features library and rules library.