Separable Fast 2D-Convolution (under construction)
Approximate fast 2D-convolution using Singular Value Decomposition (SVD).
Contents
Convolution
Convolution is a mapping of two functions (
and
) to one function (
), such that
=\int_{-\infty}^\infty f(s)g(t-s)ds.$$](SVDConv_eq40017.png)
The convolution is a weighted running average. It is a kind of multiplcation for functions. It is commutitive:

The (-) sign is
m-files
- svdConv.zip All m-files zipped.
- svdConv.m (This File)