# Difference between Auto- and Cross-correlation function

1. Dec 4, 2014

### shayaan_musta

Hello

I want to know what is the difference between Auto-Correlation function and Cross-Correlation?

I googled both but didn't find any answer which I could understand. I want you people to make me understand with quite easy language just like a spoon feeding.

And also tell me what information we get from these two?

2. Dec 4, 2014

### mathman

Auto-correlation refers to correlations between two instances within a series (or two instances of a stochastic process). Cross-correlation is about correlation between instances of two different processes.

The information tells you how strong is the relationship.

3. Dec 5, 2014

### Khashishi

Cross-correlation compares two (usually different) data series. Auto-correlation is the cross-correlation between one series and itself, so it's a special kind of cross-correlation.

4. Dec 10, 2014

### Staff: Mentor

The combination of correlation theory and noise (Gaussian or pseudo-random) was an area of study that I found captivating when I was a final year student. An autocorrelation of the output of an amplifier or a control system can reveal that system's impulse response. You can input your own noise signal, or perform this test using the small amount of noise that is naturally present at the input. In the latter case, you cause no disturbance to normal operation because you are carrying out the test without the need to input a specific signal.

Cross-correlation has endless possibilities. Suppose your job was to guard the strongroom of a bank. You could install a couple of microphones spaced well apart on the floor or walls. These microphones would pick up noise of all description, both from within the building and outside, including that of passing street traffic, and overhead aircraft. A cross-correlation of the signals from the two microphones would see most of the transitory noises smoothed over, just leaving identifiable steady spikes indicating airconditioner noise, office clocks ticking, etc. But should you notice the appearance of a new spike in the correllogram---a spike indicating a new and sustained source of sound---it could be the signature of someone tunneling nearby. Even though it might well be drowned out in amplitude by a host of other sounds, it gets revealed by the cross-correlation process.

The textbook we used was co-authored by Tom Ledwidge. https://books.google.com.au/books?id=zMtfAAAAMAAJ&focus=searchwithinvolume&q=autocorrelation

That book is not available for reading at google books, but I like this snippet:

:D