Discriminative Normalization Method for Handwritten Chinese Character Recognition

Comparing with conventional character normalization methods not taking the discriminative information into account, this paper proposes a novel normalization method -- Discriminative Normalization. Saliency regions contain most of discriminative information among similar characters. According to different types, they are enlarged in character normalization to increase their influence in recognition. As a result, discrimination power among similar characters is enhanced which is benefit to separating similar characters. The experiment on CASIA dataset shows that error rate is reduced by 9.97%. Comparing with similar character recognition without discriminative normalization, 46.0% more errors are reduced. That verifies its effectiveness.