A framework for recognizing and segmenting sign language gestures from continuous video sequence using boosted learning algorithm

The problem of vision-based sign language recognition, which is used to translate signs to English sentence, is addressed in this paper. A fully automatic system to recognize signs that starts with breaking up signs into manageable subunits is proposed. A framework for segmenting and tracking skin objects from signing videos is described. A boosting algorithm to learn a subset of weak classifiers for extracted features to combine them into a strong classifier for each sign is then applied. A joint learning strategy to share subunits across sign classes is adopted, which leads to a more efficient classification of sign gestures. Experimental results shown by the system demonstrate that the proposed approach is promising to build an effective and scalable system on real-world hand gesture recognition from continuous video sequences.