Detection and tracking of face and facial features for Recognition of Turkish Sign Language

Abstract

The accurate localization of facial features is the basic step for interpreting the facial gestures, which we use in our daily conversation to express our feelings and opinions. This research, planned to use in Turkish Sign Language Recognition systems, introduces a new method, called MISSE, whose aim is to find the facial features and to track them. The success rate of this method is approximately %92. Moreover, the test results show us, that this method can be adapted to a real-time system easily because of its speed.