Dynamic Time Warping with Python

Dynamic Time Warping (DTW) is a method to align two sequences such that they have minimum distance. For instance, two trajectories that are very similar but one of them performed in a longer time. Here is an example of my code with python. Here is my ROS package with C++ for DTW.

DTW applied two sinusoidal time series.

DTW applied two sinusoidal time series in the presence of noise
0 0 votes
Article Rating
Subscribe
Notify of
guest

This site uses Akismet to reduce spam. Learn how your comment data is processed.

3 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
trackback

[…] time. The alignment should be is such way that minimizes the distance between these two sequences. Here I have done DTW between two-time series with python. I found a good c++ library for Fast Dynamic […]

Atom
4 years ago

Hey, I think you are getting the for loop wrong because you are not including index_b!
Might be a python2 thing, I’m not sure.
Anyway I was able to reproduce the second plot in python3 with the slightly modified code below.

import matplotlib.pyplot as plt
from scipy.spatial.distance import euclidean
from fastdtw import fastdtw

start=0
end=2*np.pi
step=0.1
k=2

x1=np.arange(start,end,k*step)
x2=np.arange(start,end/k,step)

noise=np.random.uniform(start,end,len(x2))/10

y1=np.sin(x1)+1*np.sin(2*x1)+noise

y2=np.sin(k*x2)+1*np.sin(k*2*x2)
sin1=plt.plot(x1,y1)

plt.setp(sin1,color=’b’,linewidth=2.0)

sin2=plt.plot(x2,y2)
plt.setp(sin2,color=’r’,linewidth=2.0)

time_series_A=list(zip(x1,y1))
time_series_B=list(zip(x2,y2))
distance, path = fastdtw(time_series_A, time_series_B, dist=euclidean)
print(distance)
print(path)

for i,j in path:
x1=time_series_A[i][0]
y1=time_series_A[i][1]
x2=time_series_B[j][0]
y2=time_series_B[j][1]

plt.plot([x1, x2], [y1, y2], color=’k’, linestyle=’-‘, linewidth=2)
plt.show()

3
0
Would love your thoughts, please comment.x
()
x