2019 | OriginalPaper | Buchkapitel
Typing Technology of Virtual Character of Animation Based on Enneagram Personality
Autoren: Jiang-lin Luo, Mei-li Zhu, Qing-qing Wang
Verlag: Springer Berlin Heidelberg
Erschienen in:
Transactions on Edutainment XV
Abstract
Facial expression inevitably leads to facial deformation. In this paper, the advantages and disadvantages of feature extraction and recognition method are considered. A facial expression recognition method based on mixed feature fusion of nine types of personality is proposed. The texture features are extracted by discrete wavelet transform and standard orthogonal non negative matrix decomposition for a person’s facial expression image sequence, and AAM square is used. The method calculates the coordinate difference between the expression key points of the expression frame and the neutral frame in the image sequence, and extracts the geometric deformation characteristics. Then use the canonical correlation analysis (CCA) to fuse the two features, and finally use discrete HMM to classify faces.
