摘要
面部特征是实现面部表情分类与刻画面部表情强度的重要信息。提出了结合金字塔分解技术和小波矩的面部特征匹配定位方法。该方法通过小波矩实现图像信息的多尺度表征,而应用金字塔分解在金字塔图像各层之间传递信息并最终实现局部处理与全局处理之间的联系。实验结果表明,提出的方法可以在面部特征发生较大形变时依旧取得较好面部特征匹配定位结果。
Facial features contain important information for facial expression analysis and its intensity estimation.In this paper,a new facial feature detection method,which based on pyramid decomposition and wavelet moment, is proposed.It provides multi-res- olution representation of the original images and combines images＇ globe and local information to detect the facial feature.Experi- mental results show that it can accomplish the facial feature detection even if facial feature＇s deformation is large.
出处
《计算机工程与应用》
CSCD
北大核心
2007年第26期245-248,共4页
Computer Engineering and Applications
关键词
小波矩
金字塔分解
特征点匹配
wavelet moment
pyramid decomposition
feature point matching
