What is the role of 'maxsamp' argument in Dynamic Time Warping (DTW) function in MATLAB 2016b

Hi

I am having difficulty understanding the 'maxsamp' argument for the dynamic time warping 'dtw' distance function. The explanation given in the below documentation seems to be quite brief and was not able to find other relevant documents around.

https://au.mathworks.com/help/signal/ref/dtw.html

Would appreciate if someone can give an example or explanation.

Best

Baran

Answer 1

[Walter Roberson]

Fit a straight line over x and y, like

coeffs = polyfit(x, y, 1)

Now, as you go along doing the warping, ignore any sample that is more than maxsamp away from that line.

This sort of declares anything too far away from the average to be an outlier that should be ignored in calculating the warping.