Using 3D deformable template trellis to describe the movement of the lip

• Main Authors: Foo, Say Wei, Yong, Lian, Liang, Dong
• Other Authors: School of Electrical and Electronic Engineering
• Format: Conference or Workshop Item
• Language: English
• Published: 2009
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems
• Online Access: https://hdl.handle.net/10356/91394; http://hdl.handle.net/10220/6037
• Institution: Nanyang Technological University

Lip reading becomes a fascinating research area in multimedia since 1990’s. It is proven with many experiments that even small effort is made toward the incorporation of visual signal, the performance of a speech recognizer is improved compared with a purely acoustic recognizer. Such enhancement effect is especially prominent under noisy environment such as in bus station, airport, office and stock market. As a result, lip reading holds the promise to broaden the range of applicability of speech recognition to unfavorable environments. Most previous studies focus on tracking the movement of the lip during speaking. In this paper, a novel approach for this purpose is also reported. By applying 3D templates, together with probability trellis, we show that the movement of the lip at various head positions can be properly tracked.