Dynamic Time Warping (DTW). This figure exemplarily explains DTW. (A) We compute a matrix in which each entry is the Euclidean distance between datapoints from signals 1 and 2. The colour code for the various distances is shown below. (B) We use the distance matrix from (A) to develop the accumulated distance matrix D. The two red lines indicate the warping window. The warping path illustrated as blue line has to fit within the borders of the warping window. (C) The warping path explicitly states which datapoints of signal 1 align with what datapoints of signal 2. The warping path is always at least as long as the longest signal that is warped. In this example, the warping path exceeds both signals’ length, the warping path is 9 datapoints long.