关键词: 特征点跟踪, Gabor小波, 主动外观模型

Abstract: A face alignment scheme was proposed by integrating the Gabor-wavelet facial feature points tracking algorithm with the facial feature points localization method based on statistical mode AAM(Active Appearance Model). First the facial feature points were tracked with Gabor wavelet, then the result was used as the initial shape of the AAM. The local tracking errors of Gabor wavelet were corrected with the global constrained shape and texture information of AAM. Experiments prove that the method works well.

Key words: feature point tracking, Gabor wavelet, AAM(Active Appearance Model)