The recognition of finger-spelling alphabets for Chinese sign language

As a primary component of Chinese Sign Language(CSL), finger-spelling language plays an important role in deaf education and their communication. With the help of spelling rule, the system that can recognize the alphabet flow can understand and interpret CSL. So in this paper the method of effective frame extraction is proposed and used to eliminate the transition frames in the alphabet flow. And Artificial Neural Network (ANN) is introduced to recognize Chinese finger-spelling alphabet, in which a kind of fast multi-layer neural network learning algorithm-Single Parameter Dynamic Search Algorithm(SPDS) is used to learn net parameters. In addition, a recognition algorithm for finger-spelling alphabet based on multi-feature and multi-classifier is proposed and used to promote the recognition performance of alphabet. At last real-time finger-spelling language recognition system is implemented based on the above algorithm. From experiment result, it is shown that Chinese finger-spelling alphabet recognition based on multi-feature and multi-classifier outperforms its recognition based on single-classifier.