Multi-dimensional time warping based on complexity invariance and its application in sports evaluation

Complexity invariance of time series is recently considered in distance measurement, which was often neglected before in study of time series. According to related studies, this thought is more effective than the existing methods in many kinds of time series. In sports training area, there are quite a lot of requirements in motion evaluation. We decide to describe the correctness of a motion by comparing the similarity between its time series and the standard model. During the matching process, we believe the introduction of complexity invariance will improve the results of evaluation as well as the time consumption. Thus, in this paper, we use Complexity-Invariant Distance(CID) to measure the distance between multi-dimensional time series on a certain time stamp, and apply it in motion evaluation of dance training. From the results of real world practice, we obtain better performance in both accuracy and time cost.