研究优化人脸特征提取问题，针对长期以来在不贴标记点的情况下用传统的光流、Snake、可变模板等方法对纹理特征变化大的特征点不能有效跟踪，并且解决单独采用Gabor 小波系统开销大等问题，为了在人脸图像中提取准确信息，提出了人脸特征点的跟踪方法，分组采用改进的光流法和弹性图匹配的方法进行特征点跟踪。对眼睛、眉毛、上下眼皮等14 个表

情变化不大的特征点使用光流法进行跟踪，最后对变化大的嘴部8 个特征点运用Gabor 小波的弹性图匹配方法进行仿真。-Gabor wavelet research to optimize facial feature extraction problem for a long time in the case of stickers marked point changes of texture features with traditional optical flow, Snake, variable template feature points can not effectively track and solve alone overhead and large, in order to extract the accurate information in a face image, the facial feature point tracking method, grouping, improved optical flow method and elastic graph matching feature point tracking. Little change in the feature point of eyes, eyebrows, and eyelids 14 expression using the optical flow method for tracking simulation using Gabor wavelet elastic graph matching method, and finally on the changes of eight characteristic points of the mouth portion.