Transactions on Mass-Data Analysis of Images and Signals (ISSN:1868-6451)
Volume 2- Number 1 - September 2010 - Pages 55-67
Lip Tracking by Combining Simple Classifiers using AdaBoost Algorithm
M.T. Yang, Z.W. You and Y.C. Shih
National Dong-Hwa University, Taiwan
Abstract
Tracking lip contours from an image sequence is an important problem for many vision-based and bioinformatics applications. We propose a real-time method to extract the outer and inner lip contours by integrating several simple classifiers. The color/edge features and spatial-temporal consistency are exploited to make simple lip contour pixel hypothesis. The AdaBoost algorithm is adopted to combine the simple hypotheses into a strong classifier to distinguish the lip contour from non-contour pixels. A lip shape model is then deformed according to the classification scores. The experiments show that the proposed algorithm can detect both inner and outer lip contours efficiently and reliably.