how to calculate normalized cross correlations for two signals ????

hello

I hope you are doing well and fine

i have two arrays of data for two different signals

i want to compare them using cross correlations

xcorr method is used to get the correlation value and then lag value and i plot it using this code

   [tmpr lags]=normxcorr2(powersignal1,power_b_int);
    [tmpr1 lags1]=normxcorr2(powersignal1,powerT1);

my problem is that i want to do a normalized cross-correlation as it is more exact and gives better results however i do not know how to do so

is there a function in Matlab that does normalized cross-correlations calculations for different lags and return the results ??

i have been searching for a while yet i could find any

i know the formula for calculating the normalized cross-correlations but hoped for a prepared method to use right a way instead of writting the method thank you so much

Answer 1

[Yuvaraj Venkataswamy]

Check this, where x and y are two vector with length N.

if true
  Normalised_CrossCorr = (1/N)*sum((x-mean(x))*(y-mean(y)))/(sqrt(var(x)*var(y));
end

[Ed Hughes]

Hi Yuvaraj,

Thanks for your response to this. I had a follow up question - do you have to transpose any of the vectors in order to do the multiplication or would this be done element-wise (i.e., ".* "). Thanks for any guidance

Ed

[Rohan Dekate]

Hi

I implemented this for a sensor signal and a modelled signal, and got 0.847 as the answer. Correct me if I'm wrong in understanding the concept -

If the Normalized Cross Correlation is 1 it means that the two signals are matching each other and if it is 0 then they are not matching at all. Because it is normalized the answer will be between 0 and 1.

However, I have a simulink model from which I'm generating two signals as output. When I implemented the above formula I got 623700 as my answer which is not plausible at all. Can you tell me what has gone wrong?

[Vasanthi Ravindran]

Hi Yuvaraj Sir ,

Can you kindly send the formula for Normalised Cross Correlation