제목 Active Appearance Model-Based Facial Expression Recognition Using Face Symmetrical Property in Occluded Face Image
저자명

Jaehyun So, Sanghun Han, Hyun-Chool Shin, and Youngjoon Han

초   록

 This paper proposes an Active Appearance Model–based recognition of the facial expression using face
symmetrical property in occluded face image. It is very difficult to classify the occluded facial expression due to disappearance of expressional features. Unlike many papers about the recognition of the facial expression, the paper deals with the method that the expression of the occluded face is classified. To solve this problem, we use AAM
model with an half of the face features using the symmetrical property of the face. This model uses Principal Component Analysis(PCA) method for the adaptive variation of the facial shape and textures in the pre-processing time. We use an original image and a flipped image of facial expression because the model is about a half of the face. And we have used the Inverse Compositional Alignment (ICA) algorithm to fit a input face into the face model. After the AAM model fitting, the Support Vector Machine (SVM) is applied for the training and classifying of the facial expression. The efficiency of this algorithm was evaluated using Japanese Female Facial Expression(JAFFE) Database.

Key Words Facial Expression Recognition, Active Appearance Model, Principal Component Analysis, Project-Out Inverse
Compositional Alignment, Support Vector Machine
원문 수록처 International Conference on Civil, Biological and Environmental Engineering (CBEE-2013) Nov. 21-22, 2013 Bangkok (Thailand), pp 91~94
자료유형 international journal