Mean Square Error (MSE)

there are 208 images and the Dimension of each image is 256*256, now for each image I have extracted RGB value i.e 256*3 (768) then I have obtain a feature vector in fvector(v,z) the dimensions of fvector is 208*768. Now i need to calculate MSE, the MSE should be like

MSE of 1 with 1=
MSE of 1 with 2=
MSE of 1 with 3=
.
.
.
.
.
.
MSE of 1 with 208
MSE of 2 with 1
MSE of 2 with 2.
.
.
.
.
.
.
.
.
MSE of 208 with 208

How to calculate this?

[Walter Roberson]

"MSE of 1 with 3" -- what does that even mean ?

Answer 1

[Guillaume]

I'm assuming that the MSE you refer to is calculated according to the first formula on the wikipedia page. To apply this to your feature vector, use bsxfun:

mse = squeeze(sum(bsxfun(@minus, fvector, permute(fvector, [3 2 1])) .^ 2, 2))

element ij of the mse matrix is the mean square error of feature vector i with feature vector j. The matrix is obviously symmetrical.

Answer 2

[Image Analyst]

There is an immse() function in the Image Processing Toolbox that you might want to use.