Chinese characters recognition via Racah moments

With the development of the moment technology, the moment descriptors have been used in many applications in the field of object recognition. In recent years, discrete orthogonal moments have been studied and proposed in moment applications as features to characterize image shapes. In this paper, a set of Racah moment based feature vectors is proposed for a Chinese character recognition system. The experimental results demonstrate that the proposed Racah moment feature vectors can effectively recognize Chinese characters, including the Chinese characters that have very similar structures and shapes.