Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
Drexel University or The Math Forum.


Nunun
Posts:
1
Registered:
10/17/09


Singular value decomposition in noise reduction
Posted:
Oct 18, 2009 7:11 AM


SVD, given matrix A. A = UE(V^T)
I understood that the singular values (diagonal E) are the noise and need to be reduced to obtain a clearer image. But how does reducing the singular values actually affect the overall picture? Is it because of the change in A, U and V?
From my understanding, the bigger the singular values are, the more noise there are in the image. Hence, after obtaining diagonal E from A, we reduce the E and then compute A, is it? Which of these (A, U, V) actually makes the new image clearer?



