Fig. 4
From: Autonomous modeling of repetitive movement for rehabilitation exercise monitoring
Comparison of the data structure used in the original DTW and Circular DTW. Each black square represents a matching point between one sample in a template sequence and one sample in an unsegmented smoothed phase sequence (output of Algorithm 2). The original DTW is impractical as the proper number of rows is unknown before the calculation. The Circular DTW lets one finite template sequence be reused for multiple rounds as needed by bridging the first and the last row of the table