Gaze tracking and its application to video coding for sign language.

Abstract

Sign language communication via videotelephone has demanding
\nvisual quality requirements. In order to optimise video coding for
\nsign language it is necessary to quantify the importance of areas of
\nthe video scene. Eye movements of deaf users are tracked whilst
\nwatching a sign language video sequence. The results indicate that
\nthe gaze tends to concentrate on the face region with occasional
\nexcursions (saccades). The implications of these results for
\nprioritised coding of sign language video sequences are discussed.