A method for recognizing facial expression intensity based on facial muscle variations
Yukun Zhang,Zixiang Fei,Xia Li,Wenju Zhou,Minrui Fei
DOI: https://doi.org/10.1007/s11042-024-19779-4
IF: 2.577
2024-07-14
Multimedia Tools and Applications
Abstract:Expression intensity recognition is a research problem in the field of computer vision and pattern recognition, which can be understood as the salience, or intensity, of an expression. Existing techniques for expression intensity recognition find it difficult to capture the texture features of expressions, which results in low recognition accuracy. Therefore, this paper proposes a method for recognizing expression intensity based on facial muscle changes to address this issue. This method captures texture motion information of facial muscles using optical flow, determining the motion regions and static regions of facial texture based on optical flow magnitude. The movement areas of facial expressions are set as salient regions, and the intensity of emotions is classified based on the variation rate of muscles within these regions. The proposed algorithm exhibits good recognition accuracy on publicly available datasets such as CK + , MMI, and datasets collected from real scenarios, including those involving elderly individuals. The average recognition accuracy reaches 82.7%. The proposed algorithm demonstrates a strong ability to recognize the intensity of facial expressions. This study provides technical support for the development of a real-time mood monitoring system.
computer science, information systems, theory & methods,engineering, electrical & electronic, software engineering