finding MSE and MAE for RGB image

i have a equation below please tell how to find MSE and MAE for RGB image

http://imgur.com/94kpN

Answer 1

[Walter Roberson]

To within a constant factor that might depend upon image size,

MSE = sum( (FirstImage(:) - SecondImage(:)).^2 );

[FIR]

Walter then for MAE

[FIR]

my image is peppers.png

[Image Analyst]

It's just the absolute value of the differences! Be sure to cast to double to allow negative numbers. Subtracting uint8's will clip to zero. Is this homework?

[Walter Roberson]

Good point about casting to doubles. I should have said

MSE = sum( (double(FirstImage(:)) - double(SecondImage(:))).^2 );

[FIR]

can u please tell what is the difference in notation of 1 and 2,both equation are sme but notation is different

[Walter Roberson]

As the document indicates, |x| with a subscript of 2 corresponds to the L2 normal, which is the Euclidean Norm, also known as Euclidean distance. The Euclidean distance is like sqrt( (x1-x2)^2 + (y1-y2)^2) and so on. Notice that in the formula given, they square this value -- that is what the superscript 2 is about, squaring the result of the Euclidean distance. The squaring cancels out the sqrt(), leaving (x1-x2)^2 + (y1-y2)^2

The |x| with a subscript 1 is the L1 norm, also known as the taxicab distance or the Manhattan norm. This corresponds to x1-x2 + y1-y2 with no sqrt(). Notice though that in the formula they do not square this, leaving x1-x2 + y1-y2, abs(x1-x2) + abs(y1-y2)

[FIR]

so walter can u tell how to write eq for MAE PLEASE

[Walter Roberson]

Like MSE except using abs() instead of .^2