Forums

normalized cross correlation

Started by vondykes February 7, 2008
Hello

I have done a cross correlation but still have aproblem of getting wrong
result when two signals is differnent in  length. I solved the problem by
padding zeros to the shorter one to have equal length with another signal,
but the problem still exist. I have found out that I need to normalized the
cross correlation which I don't know how to do so. I have found out an
equation, 

rxy(k)=E[x(n)-x']*[y(n+k)-y']/[E[x(n)-x']^2*[y(n)-y']^2]^1/2

provided that I have signal x and y(the shorter one) and "'" is the mean
of signal.
E is the summation from k=1 to N

Sorry that it looks difficult be cause I don't know how to type such
equation.

I'm not sure whether this equation is right and I also got confused with
the length N too. Is this the length of signal x or y? So if anybody knows
please help. I would be really appreciate.