How to find starting point of one signal in another?

I have 2 signals that have the same sampling frequency (100 Hz), but one began recording before the other. I'm trying to find the size of this delay (# of indices/sample points) so I can align the signals.

I have tried using the 'alignsignals' function, but the delay it gives me is clearly to big, and when I manually try to find a matching point in the two signals and subtract one from the other to find the delay, it still doesn't seem to give me an accurate delay. I've also tried using xcorr but it's giving me the same delay as the alignsignals function whic again, seems to big.

Does anyone have any suggestions or know what I might be doing wrong? Signals attached.

Thank you.

[Adam Danz]

Am I correct in thinking that the noisy segment between x=0 to x=~2.25 for the blue data are noise and the data begins at ~x=2.25 (same logic for the red data)?

If that interpretation is correct, is it expected that the noise for the red data are several orders of magnitude larger than the noise in the blue data?

If that interpretation is incorrect, please explain what indicates the start (or end?) of the signal.

clf()
subplot(2,1,1)
plot(task_left_foot,'b-')
subplot(2,1,2)
plot(vicon_left_foot, 'r-')

[Maria Y]

Hi Adam,

Thanks for your response. I am trying to find where in the red signal the blue signal is located. x=0 to x=~2.25 is the entire signal I'm looking for.

If you zoom into the red signal from ~x=1 to the end, you'll see that visually the signals look the same (which they should because they are both foot position data from the same system recording the exact same events at the same time, the blue signal just started slightly after the red). So visually I can see they start lining up around ~x=1 but I'm trying to find the exact indice in which this happens.

[Adam Danz]

"you'll see that visually the signals look the same "

I admitt that I haven't zoomed in to the data but when I plot the data on the same axes so we get a sense of scale, it's hard to imagine that the signals will look the same, even if I imagine that one is shifted relative to the other. As I mentioned before, the vicon data are magnitudes larger than the task data.

clf()
plot(task_left_foot,'b-')
hold on
plot(vicon_left_foot, 'r-')

They are both labeled "left_foot". Could it be that the wrong data were attached?

Answer 1

[Pablo Fernandez]

Try the findsignal function:

load('task_left_foot.mat');
load('vicon_left_foot.mat');
findsignal(vicon_left_foot,task_left_foot)

[Maria Y]

Thank you Pablo. Sorry I meant to follow up on this post - I realized there were some missing data points in one of the signals which is why functions such as findsignal weren't working. Figured it out now.

Thanks again.

Maria

[Pablo Fernandez]

Good to hear you solved your problem. I did notice that alignsignals gives a different, less correct answer.