Should the best cross correlation value be the larger or smaller?

If i use the xcorr function to get the max cross correlation of signal s1 compared to s2 and then i compare the cross correlation of a different signal to signal s2 and get the max of that. Should the signal that is a better match be the larger or the smaller? I had assumed the larger would be the better because its more correlated. But having tried this out the signal with the lower max cross correlation looks like a better fit?

[Ronan]

So this doesnt make sense. If you want to define how well they match you can use something like the sum of squares. But xcorr looks for where on the signal they closest match. So you cant say something like signal A is better correlated to signal x than signal B is correlated to signal x. The "goodness" of correlation is where one signal aligns best with another so its out of context to compare to another signal.

Answer 1

[Mahdiyar]

Hi Depend on what you want to do with the signals s1, s2,.... Sometimes you need to have some completely unique signals which have the least similarity together as much as possible. In this case it would be better to have the low cross correlation.

If you need to use signals with the most similarity, the higher cross correlation is more desirable.