A new radical-based approach to online handwritten Chinese character recognition

This paper proposes a new radical-based approach for online handwritten chinese character recognition. The approach is novel in three respects: statistical classification of radicals, over-segmentation of characters into candidate radicals, and lexicon-driven recognition of characters. Currently, we have applied the approach to Chinese characters of left-right structure and are extending to other structures. Preliminary results on a sample set of 4,284 characters consisting of 1,118 radicals demonstrate the superiority of the proposed approach.