The visual speech recognition (lipreading) subsystem

• Block diagram of the visual speech recognition subsystem:

• • Sometimes, also face detection is needed as first step, to obtain the speaker’s face image sequence

Department of Informatics

Aristotle University of Thessaloniki

Speaker’s

face image

sequence

Mouth localization => ROI

Facial

feature

extraction

Mouth image

Facial features analysis

Lipreading

Visual speechreading result

Facial

feature vector

Previous slide

Next slide

Back to first slide

View graphic version