Coarse-DTW: Exploiting Sparsity in Gesture Time Series

Dynamic Time Warping (DTW) is considered as a robust measure to compare numerical time series when some time elasticity is required. Even though its initial formulation can be slow, extensive research has been conducted to speed up the calculations. However, those optimizations are not always available for multidimensional time series. In this paper, we focus on time series describing gesture movement, all of which are multidimensional. Our approch propose to speed up the processing by 1. adaptively downsampling the time series into sparse time series and 2. generalizing DTW into a version exploiting sparsity. Furthermore, the downsampling algorithm doesn’t need to know the whole timeseries to function, making it a good candidate for streaming applications such as real-time gesture recognition.