PAD Model Based Facial Expression Analysis

详细信息    查看全文

• 作者：Jie Cao ; Hong Wang ; Po Hu ; Junwei Miao
• 刊名：Lecture Notes in Computer Science
• 出版年：2008
• 出版时间：2008
• 年：2008
• 卷：5359
• 期：1
• 页码：450-459
• 全文大小：813.9 KB

文摘

The validity of PAD (Pleasure-Arousal-Dominance) theory in vision area and the feasibilityT on PAD based models for facial expression analysis are discussed in this paper. Three new models based on PAD theory are proposed and their feasibility is verified by experiments on Cohn-Kanade dataset and PAD dataset which is collected from well-designed psychological experiments. After combining Gabor feature and SVM (Support Vector Machine), the result can be further improved. Compared with the basic expression models, our experiments show that the predominance of PAD based model is that it can represent almost any states of expression. Finally, our preliminary experiments show that distinguishing different grades of the same expression is promising by our models.