High frequency noise in GRF values from the lab. Is this normal?

[Daniel Pina]

Hello,

I'm still struggling with high-frequency noise in my SO results from another thread.
(see here: viewtopicPhpbb.php?f=91&t=8059&p=0&star ... c8b99c16e0 )

I was wondering if the problem might be because all my GRF files also have noise in them, and if applying a low-pass filter could somehow solve these problems.

Here's an example of the GRF I have for one of my lab results, it represents a step during gait at "normal" speed:

grf_with_noise.png (19.37 KiB) Viewed 730 times

And here is a "zoomed" graphic of the example above:

grf_with_noise_zoom.png (20.66 KiB) Viewed 730 times

Could this be the reason for me getting high-frequency noise in my SO results?




Thank you in advance.

[Dimitar Stanev]

It is possible! You can try to smooth (pre-process) the GRFs before feeding to SO and see if things improve.

[Daniel Pina]

Just a quick follow-up:

Filtering the GRF filtes worked!

I had some problems at first because I forgot that adding a single-pass filter would result in a delay of the resulting signal, and this skewed my results.

I had to apply a zero-phase filter (2nd pass inverted) to the initial results to get the intended results.

In Matlab or Octave this means using the "filtfilt" function instead of "filter".


These are my new results for one mocap data for walking that I have:

solved_own_results - cut.png (154.86 KiB) Viewed 672 times

And these are results I've been as reference from here :

solved_reference - cut.png (217.41 KiB) Viewed 672 times