A Real-Time Large Vocabulary Recognition System for Chinese Sign Language

摘要

The major challenge that faces Sign Language recognition now is to develop methods that will scale well with increasing vocabulary size. In this paper, a real-time system designed for recognizing Chinese Sign Language (CSL) signs with a 5100 sign vocabulary is presented. The raw data are collected from two CyberGlove and a 3-D tracker. An algorithm based on geometrical analysis for purpose of extracting invariant feature to signer position is proposed. Then the worked data are presented as input to Hidden Markov Models (HMMs) for recognition. To improve recognition performance, some useful new ideas are proposed in design and implementation, including modifying the transferring probability, clustering the Gaussians and fast matching algorithm. Experiments show that techniques proposed in this paper are efficient on either recognition speed or recognition performance.