Wavelet係数を用いた時系列の標準化手法 : 文楽人形の演技動作時系列の標準化への応用

概要

論文の詳細を見る
Observe same phenomena K times, then one obtain K sample paths x_1(・), x_2(・), …, x_K(・)of the time series. The graphs of these K sample paths are similar, but they do not same exactly. Although the state x_1(t_1)at time t_1 of the 1 st sample path and the state x_2(t_2) at time t_2 of the 2 nd sample path mean same events, t_1≐̸t_2 and x_1(t_1)≐̸x_2(t_2)in general. Time normalization between two sample paths x_1(・)and x_2(・)is the process of finding the time t_1 of the 1 st sample and the time t_2 of the 2 nd sample path, such that x_1(t_1) and x_2(t_2)show same events. Consider graphs of two sample paths. If the local form of the graph of the sample x_1 at time t_1 is similar to the one of the sample x_2 at time t_2, we can think that x_1(t_1)and x_2(t_2)mean same events. In order to evaluate the similarity of the local forms of the graphs, the wavelet coefficients of the sample paths are used. If one know x_1(t_1), x_2(t_2), …, x_K(t_K)mean same events, one can normalize these time fluctuations and obtain new average time series (x^^-)(t^^-) where (t^^-)=(t_1+…+t_K)/K and (x^^-)(t^^-)=(x_1(t_1)+…+x_K(t_K))/K. This time normalization method is applied to action time series of a Bunraku puppet, for generation of humanoid robots' actions with fertile emotions.
社団法人日本機械学会の論文
2000-06-25