How can i implement least squares of the difference between 2 matrices?

Hi,

I'm trying to get the least squares of the difference between 2 matrices

argmin |g-u| both g and u are matrices of size [280,307,191]

any help?

Answer 1

[Iman Ansari]

Hi

Error= g-u;
MSE=mean(Error(:).^2)

[Rawan hamdi]

what about the term arg min " |g-u| "

 "||g-u||"  is the least squares

Thank you

[Iman Ansari]

I'm not sure:

Error= g-u;
SquersError=Error.^2;
LeastSquersError=min(SquersError(:));
Argmin=Error(SquersError==LeastSquersError);

[Image Analyst]

Explain in words what "argmin" means to you. Then we can try to answer.

[Rawan hamdi]

what i understand is that it's the argument of the minimum,i want to optimize this equation :

 ||g-u|| where g and u are both images of same size .

[Iman Ansari]

For these u and g what is the answer?

u=[211   103   93;  256   239   16;  96   110   216];
g=[147   38   131;  11    228  112;  158   192   41];
Error= g-u;
SquersError=Error.^2;
LeastSquersError=min(SquersError(:));
Argmin=abs(Error(SquersError==LeastSquersError))

Is 11 the result you want?

[Rawan hamdi]

i really have no idea what the constrained least squares means! but all i understood is that the term

|| g-u || is  the data fidelity term between these 2 images(matrices) and the whole equation 
x = ||g-u ||.^2 is the constrained least squares
and i'm trying to know what the result should be