An on-line Chinese character recognition for personal communicator

An on-line Chinese character recognition (OLCCR) based all subword model for a personal communicator is presented. This OLCCR has three advantages such as short recognition time, high recognition rate and a small kernel memory size. The average rate of correct recognition is 99.1% within the top ten priorities of the recognition results. The average recognition time of one character is 0.415 second with the 486-33 personal computer. The occupied memory size of the program kernel, including the algorithm and template, is 480 Kbytes. The personal communicator would be intuitive and easy to use by combining it with the OLCCR that we developed.