摘要
提出了一种基于特征点跟踪的面部表情强度度量方法.该方法在L-K光流算法的基础上,应用修正特征点跟踪提取面部特征信息,通过非线性降维从高维特征点轨迹中抽区一维的表情强度,最后使用SVM建立表情模型和强度模型,对高兴表情进行强度等级的归类.针对高兴表情进行实验,引入表情强度等级作为衡量标准,实验结果证明了该方法的有效性.
A method of measuring the expression intensity based on facial feature point tracking is presented, which is based on the L-K optical flow algorithm and uses the amended feature point tracking to track facial feature points. A nonlinear dimensionality reduction method is used to automatically generate the spectrum of expression intensity by extracting 1-D manifold from high dimensional feature point trajectories. Support Vector Machine is used to build the expression model and intensity model to classify the facial expressions and classify the intensity grade of happy expression. The experiment is done on the happy expression, and the intensity grade is introduced as measurement standard. Experiment results validate the effectiveness of this facial expression analysis method.
出处
《重庆工学院学报(自然科学版)》
2008年第9期176-180,共5页
Journal of Chongqing Institute of Technology
基金
湖南省教育厅科研基金资助项目(06C219)
关键词
表情强度
特征点跟踪
L—K光流
SVM
expression intensity
feature point tracking
L-K optical flow
Support Vector Machine