作者: J. Luettin , N.A. Thacker , S.W. Beet
DOI: 10.1109/ICSLP.1996.607030
关键词:
摘要: This paper describes a new approach for speaker identification based on lipreading. Visual features are extracted from image sequences of the talking face and consist shape parameters which describe lip boundary intensity grey-level distribution mouth area. Intensity information is principal component analysis using eigenspaces deform with model. The account both, speech dependent information. We built spatio-temporal models these features, HMMs mixtures Gaussians. Promising results were obtained text independent tests performed small video database.