|
"dhrub " <dhruvkumar037@gmail.com> wrote in message <k8795b$ho4$1@newscl01ah.mathworks.com>...
> what is the difference between wrcoef2 and appcoef2/detcoef2 in matlab ?
>
>
> I mean in both cases, we get coefficients.
> what is the difference then...
Only in appcoef2 and detcoef2 do you actually get approximation or wavelet coefficients. With wrcoef2 you get a projection onto the subspace spanned by the particular wavelet or scaling function. There is a big difference between the coefficients and the projection. For one thing, the projection is the same size as the original image.
I'll give you a simple example with vectors in R^2 to illustrate the difference.
Assume you have an orthonormal basis for R^2: [1/sqrt(2) 1/sqrt(2)]', [1/sqrt(2) -1/sqrt(2)]'
Now take an arbitrary vector, [3 2]'. The expansion coefficients for [3 2]' in the basis are:
5/sqrt(2)
1/sqrt(2)
The orthogonal project of [3 2]' onto the space spanned by [1/sqrt(2) 1/sqrt(2)]' is
5/sqrt(2)*[1/sqrt(2) 1/sqrt(2)]'
2.5
2.5
The difference between appcoef2/detcoef2 and wrcoef2 is analogous to the difference between 5/sqrt(2) (the expansion coefficient) and [2.5 2.5], the projection onto the subspace spanned by [1/sqrt(2) 1/sqrt(2)]'
|