AVSP 2003 - International Conference on Audio-Visual Speech Processing

September 4-7, 2003
St. Jorioz, France

A System for Automatic Lip Reading

I. Shdaifat (1), R. Grigat (1), Detlev Langmann (2)

(1) TU Hamburg Harburg, Vision Systems, Hamburg, Germany
(2) Philips Semiconductors GmbH, Hamburg, Germany

In this paper, we present our approach of face and lip detection, lip modeling, and tracking. A new lip model based on B´ezier Curves is used to capture the dynamics of the lips effi- ciently. The model is defined only through few points which are modeled using the Active Shape Model (ASM). Accurate detection of lip details is implemented using multiple independent feature templates. The method detects tracks and model the lips online and robustly. The results of the lip detection are presented based on our collected data for German language. We also describe also some of the current and future work on audio video integration.


Full Paper

Bibliographic reference.  Shdaifat, I. / Grigat, R. / Langmann, Detlev (2003): "A system for automatic lip reading", In AVSP 2003, 121-126.