Research And Development Of Thai Handwritten Recognition Of Embedded System

With on-line handwritten recognition technology going to productive phase, a lot of invigorative applications of on-line handwritten recognition for many language appeared in embedded system. But Thai characters have special structure and shape, many of them have similar shape, and handwritten characters in the system have complex distortion. Therefore, handwritten Thai character recognition is one difficult task. On the other hand, the performance of CPU in embedded system is frequently slow, and don't have enough capability in dealing with complex operation. In order to approving implement Thai on-line handwritten recognition technology in embedded system, we must design the handwritten system based on embedded system condition and simplify and optimize the arithmetic.This paper proceeds from on-line handwritten recognition theory and analysis of the Thai characters. And select feature extraction method to represent the features of handwritten Thai characters effectively and absorb the distortions of handwritten. This paper designed the important stage in the recognition system based on embedded system, include preprocess, character separating and recognition stage. Finally, we design the Thai on-line handwritten recognition that have three stage classifier, and can be used in low-end embedded system.The results of experiment show that the effects of Thai character recognition that designed by this paper are observably better than those produced by current method. The method and arithmetic proposed by this paper can be used in the Thai characters recognition field.Thai on-line handwritten recognition is a simple method to input Thai character ,it is popular to recognize the continuous handwritten character. In order to promote the system, the better method and arithmetic will be study in future.