03-25-2009 11:30 AM
Hi, any signal processing experts out there that can help me out here?
Basically all i want is to get the time difference of the signals.
03-25-2009 03:08 PM
Aaargh... I still can't can't get any consistent data!
I tried using the threshold to detect the difference in time but due to the inconsistent pattern of the sound captured, the values do not match half the time. Increasing the hysterisis does not work either. Taking the location of the maximum peak works half the time only as you can see from the sample signals i posted, the vary a little from one sensor to another.
As for the cross correlation technique, i'm stuck with some weird data which i haven't a clue on how to process it.
Any other technique you guys know that could assist me in my predicament?
Help.....!
I think this time i'll try taking the time difference between each signal when they exceed the standard deviation value i set.
03-25-2009 03:36 PM
03-26-2009 12:28 AM
Hmm i did try to do some filtering but since there is no one particular frequency that my system is looking for, my filter actually filters part of the actual sound as well.
However the AIC time flight technique you recommended does look quite promising.
"Algorithm for AIC. The onset is the point having the minimum AIC value. For the seismogram x[k] (with k=1, 2…N) of length N, the AIC value is defined as
AIC(k) = k * log{var(x[1,k])} + (N - k -1) * log{var(x[k + 1,N])}
where k ranges through all the seismogram samples."
So if I was to implement this algorithm, i just key in the formula into the matchscript?
03-26-2009 11:21 AM
Be aware that the calculation is non trivial. The variance is calculated for each window on both sides. But at least try that. If not fast enough, maybe write it in C and called it using call library function node.
Good luck.
07-21-2009 03:18 AM
Hi Carefree!
I need to calculate the time delay of two signals too... And i can' to it... Did you done it? Can you post your solution?
Thanks: Kolos