EM-in-M: Analyze and Synthesize Emotion in Motion
We have been researching the relationship between human motion and emotion. In this paper, our purpose is to extract motion features specific to each emotion. We propose a new approach for motion data analysis, which applies the higher order Singular Value Decomposition(HOSVD) direct to the motion data and the wavelet analysis to the synthesized data with SVD. The HOSVD models the mapping between persons and emotions. The model can synthesize a complete data acting with each emotion for a given new person. The wavelet analysis extracts each motion feature from the synthesized data for each emotion. Some experimental results using motion capture data for “gait” action and 6 emotions – “angry, joy, sad and so on” show that our method can synthesize novel gait motions for a person by using the extracted motion elements and can extract some features specific to each emotion.
KeywordsJoint Angle Wavelet Analysis Motion Data Human Motion Motion Feature
Unable to display preview. Download preview PDF.
- 2.Shiratori, T., Nakazawa, A., Ikeuchi, K.: Detecting Dance Motion Structure through Music Analysis. IEEE ICAFGR, 857–862 (2004)Google Scholar
- 3.Kobayashi, Y., Ohya, J., Zhang, Z.: Cognitive bridge between haptic impressions and texture images for subjective image retrieval. In: Proc. of IEEE ICME, pp. 2239–2242 (2004)Google Scholar
- 4.Brand, M., Hertzmann, A.: Style machines. In: Proc. of ACM SIGGRAPH, pp. 183–192 (2000)Google Scholar
- 5.Sundaresan, A., Chowdhury, A.R., Chellappa, R.: A hidden markov model based framework for recognition of humans from gait sequences. In: Proc. of IEEE ICIP, vol. 2, pp. 93–96 (2003)Google Scholar
- 6.Vasilescu, M.A.O.: Human Signatures: Analysis, Synthesis, Recognition. In: Proc. of ICPR, vol. 3, pp. 456–460 (2002)Google Scholar
- 7.Daubechies, I.: Ten Lectures On Wavelets. Society for Industrial and Applied Mathematics (1995)Google Scholar