Screenshot from DTW.

Dynamic Time Wrapping


The purpose of Dynamic Time Warping (DTW) is to compare two different length sequences of values and allow the evaluation of an error measurement on the basis of the match between the two sequences.

Dynamic Time Warping tries to synchronize two motions by warping their time domains, which makes the motions faster or slower. Matching of the synchronized motions estimates similarities between them, which can be a base for nearest neighbor’s classification.

DTW transform is used for the reduced pose spaces by Principal Component Analysis(PCA), DTW can also be applied to the sequence of binary relation motion features, which indicates defined relationships between body parts

Features :

  • Using Cimage - To plot the DTW alanysis in X/Y dimension.
  • Cost matrix which contains the similarities between every pair of poses of compared motions
  • Synchronization is determined by the monotonic path connecting starting and ending points of the cost matrix with the lowest accumulated cost. The cost of the path found estimates motions dissimilarity
  • Currently compares only one motion along the time.space dimension.

External Links :

  1. Dynamic Time Wrapping Simple Explanation
  2. Dynamic Time Wrapping Slides (Graphical)
  3. Temporal Alignment Of Human Motion
  4. Dynamic Time Wrapping In Gait Classification Of Motion Capture Data

Video :


Source On GitHub (Coming Soon)

Download Binaries(Windows)