A High Speed Word Matching Algorithm for Handwritten Chinese Character Recognition

A new high speed word matching algorithm for handwritten Chinese character recomition is presented. A continuous string without delimiting space is recognized in real time by using this algorithm. Errors and rejects of an optical character reader arc corrected to meaningful characters with the help of a dictionary. This algorithm uses a finite state automaton (FSA); the FSA's transition table is condensed by the representation device of the FSA to reduce the processing time of table generation and state transitions. Words can be extracted from any position in the continuous string. The recognition process is 6.8 times faster than a conventional automaton-type algorithm. The algorithm was run on 5,032 handwritten Japanese sample addresses. In the experiment, 96.6 percent of the recognition errors and rejects wcrc corrected in real time.