MULTISTAGE PRE-CANDIDATE SELECTION IN HANDWRITTEN CHINESE CHARACTER-RECOGNITION SYSTEMS

This paper presents a model for reducing the processing time needed fo r recognition of handwritten Chinese characters via multi-stage pre-ca ndidate selection. The features used in pre-candidate selection are or dered according to the reduction rates evaluated from a set of trainin g characters. A multi-stage pre-candidate selection module is activate d for each input character. In each stage, we use a single feature com puted from the input character to eliminate impossible categories of c haracters. When the number of stages increases, the pre-candidate sele ction time will increase, while the candidate selection time will decr ease. The total execution time will decrease when the increase in pre- candidate selection time is less than the decrease in candidate select ion time. The total execution time can be predicted from the set of tr aining characters. The number of stages of the pre-candidate selection module is selected to be that which results in the minimum total exec ution time. Experiments are performed on the CCL/HCCR1 database with 5 401 categories of handwritten Chinese characters. The experimental res ults show that the proposed model can reduce the total execution time significantly without decreasing the precision of the candidate select ion module.