论文标题
使用光流算法分析面部特征变形
The Analysis of Facial Feature Deformation using Optical Flow Algorithm
论文作者
论文摘要
面部特征根据预期的面部表达而变形。特定的面部特征与特定的面部表达相关，即快乐意味着口腔的变形。本文通过使用光流算法进行了每个面部表达的面部特征变形的研究，并分为三个不同的感兴趣区域。面部特征的变形显示面部表达与面部表达之间的关系。基于实验，除了快乐以外，所有表达式中的眼睛和口腔变形都很明显。对于快乐的表情，脸颊和嘴是重要的区域。这项工作还表明，不同面部特征的强度在识别不同面部表达强度的方式方面有所不同。所有表达式中的最大幅度均显示为令人惊讶的表达，即9x10-4。虽然最小幅度由嘴巴显示为0.4x10-4的愤怒表达。
Facial features deformed according to the intended facial expression. Specific facial features are associated with specific facial expression, i.e. happy means the deformation of mouth. This paper presents the study of facial feature deformation for each facial expression by using an optical flow algorithm and segmented into three different regions of interest. The deformation of facial features shows the relation between facial the and facial expression. Based on the experiments, the deformations of eye and mouth are significant in all expressions except happy. For happy expression, cheeks and mouths are the significant regions. This work also suggests that different facial features' intensity varies in the way that they contribute to the recognition of the different facial expression intensity. The maximum magnitude across all expressions is shown by the mouth for surprise expression which is 9x10-4. While the minimum magnitude is shown by the mouth for angry expression which is 0.4x10-4.
