Hello all,
I have two complex signals and each signal has 2500 complex values.
What is the best to measure the similarity between these two signals?
I already applied non-linear cross correlation and I am looking for different mean.
Cheers,
This is a very cool question. Though I think we need a little bit of clarification.
So is this something like impedance values from two parts of a circuit where we expect that there might be a relationship, or is this more like results from two radio telescopes pointed in opposite directions where we hope but do not expect to find a specific correlation?
By similarity are you meaning correlation, or some similarity index?
Or is this looking for how different the mean of one signal is from the mean of the other signal?
Well, I am looking to compare one signal against a reference signal. Both signals are 2-dimensional complex values obtained by applying double random phase encoding technique.
Basically, two signals are from the same source (lets say one image) but one of the signals is distorted by adding some noises or manipulated a few bits of the original image. I am interested in checking to see weather two signals are still similar or not.
Cheers
I also thought about the same thing and I applied kind of normalized mean square error. It seems that it is working.
Perhaps the average absolute deviation (L1 norm) will be even more useful? Euclidean distance (L2 norm) is definitively more sensitive to outliers.
- Badges
- Science topic
- Similar topics
- Psychometrics
- Measurement