A Sign Language Recognition Based on Tensor

The hand gesture recognition plays a key role in many appealing applications. The sign language recognition is one of the important applications of hand gesture recognition. The existing methods on sign language recognition are limited to certain view. In this paper, we use tensor subspace analysis to model a multi-view hand gesture to recognize 26 manual alphabetical letters. In our experiment, each hand gesture is captured from 5 different views. Two experiments are conducted on gray-scale images and binary images, respectively. The results show the proposed method has a good performance on multi-view.