Discover MakerZone

MATLAB and Simulink resources for Arduino, LEGO, and Raspberry Pi

Learn more

Discover what MATLAB® can do for your career.

Opportunities for recent engineering grads.

Apply Today

To resolve issues starting MATLAB on Mac OS X 10.10 (Yosemite) visit: http://www.mathworks.com/matlabcentral/answers/159016

Xcorr2 function

Asked by Padma on 11 Sep 2011

hello,can anybody help me how to go about using the xcorr2 function to find the cross correlation between 2 images? i need a single value of correlation between a query image and an image in the database. i have used the followin: crosscorr1=max(abs(xcorr2(double(x(:,:,1)),double(y(:,:,1))))); where x and y are d 2 images. i get a single row matrix as the output.

0 Comments

Padma

Products

No products are associated with this question.

2 Answers

Answer by Wayne King on 11 Sep 2011
Accepted answer

Hi, That's because you are using max() on a matrix, which returns by default the maximum values for each column. One thing you can do is to call max() on the initial max() output.

x = randn(10,10);
y = randn(10,10);
max(max(abs(xcorr2(x,y))))

Note however that xcorr2 is not a normalized cross correlation.

Wayne

1 Comment

Padma on 11 Sep 2011

thanks!!!
but how is it different from normxcorr2??? will make a difference if i use just xcorr2?

Wayne King
Answer by Wayne King on 11 Sep 2011

The difference is that xcorr2 depends on the scale of the values in the input matrices, while normxcorr2 normalizes the cross correlation values to lie in the interval [-1,1]. Consequently, you can interpret the output of normxcorr2 more easily. If I tell the maximum value of the cross correlation between two matrices using xcorr2 is 200, you have no way of saying whether that indicates a strong linear association or not. All you know is that the correlation is in a certain direction. On the other hand if I tell you the maximum normalized cross correlation is 0.9. That gives you information about the strength of the cross correlation.

1 Comment

Padma on 14 Sep 2011

hey,thank u!!

Wayne King

Contact us