Efficient Tracking Method to Make a Real Time Sign Language Recognition System

摘要

In the field of automatic treatment of natural languages, the analysis and the exploitation of each statement in sign language (SL) have a great importance. In fact, the own specificities of SL such as the simultaneity of many parameters, the significant role of the facial expression, the use of space to structure the statement, as well as the technical specificities, such as the change lightening and the presence of occlusion in the space of one-sighted-capture, have a deep effect on tracking the different parts of the body. In this paper, we propose an empiric method of tracking adapted to the specificities of SL that we use to elaborate a real time recognition system based on a prediction approach.