Fast Motion Retrieval with the Distance Input Space

Accessing and querying large motion databases can be a time consuming operation since motions are by essence big multivariate time series. However, with the help of good dimensionality reduction methods, this problem can be alleviated. Yet, it appears that the intitial motion representation space if of great importance with regards to the final performances. This paper advocates the idea that the distance representation space (i.e. ensemble of joint-to-joint distances of the skeleton), though bigger, can yield a powerful representation of the motion. In order to validate this idea, comparisons with rotation and position based representations are experimented in a motion retrieval task performed with the HDM05 motion database. The final resulting retrieval algorithm is very fast with satisfying performances.