Facial Expressions Recognition Based on Cognition and Mapped Binary Patterns

• Published in: IEEE access Vol. 6; pp. 18795 - 18803
• Main Authors: Qi, Chao, Li, Min, Wang, Qiushi, Zhang, Huiquan, Xing, Jinling, Gao, Zhifan, Zhang, Huailing
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01-01-2018; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Cognition; Cognition & reasoning; Datasets; Education; emotion model; Emotions; Face; Face recognition; Facial expressions recognition; Facial features; Feature extraction; human cognition; local binary patterns; machine learning; Principal component analysis; Support vector machines; Three dimensional models
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2018.2816044

In this paper, a new expression recognition approach is presented based on cognition and mapped binary patterns. At first, the approach is based on the LBP operator to extract the facial contours. Secondly, the establishment of pseudo-3-D model is used to segment the face area into six facial expression sub-regions. In this context, the sub-regions and the global facial expression images use the mapped LBP method for feature extraction, and then use two classifications which are the support vector machine and softmax with two kinds of emotion classification models the basic emotion model and the circumplex emotion model. At last, we perform a comparative experiment on the expansion of the Cohn-Kanade (CK+) facial expression data set and the test data sets collected from ten volunteers. The experimental results show that the method can effectively remove the confounding factors in the image. And the result of using the circumplex emotion model is obviously better than the traditional emotional model. By referring to relevant studies of human cognition, we verified that eyes and mouth express more emotion.