Wiener filter

A regularized inverse filter that damps unreliable frequencies toward zero instead of dividing by them.

The Wiener filter estimates $\hat F(u,v) = \dfrac{H^*(u,v)}{|H(u,v)|^2+K}\,G(u,v)$ , where $K\ge0$ stabilizes the division: as $K\to0$ it reduces to inverse filtering exactly, and as $K$ grows it damps an unreliable (small- $|H(u,v)|$ ) frequency's estimate toward $0$ instead of amplifying its noise. The frequency-domain special case of ridge-regularized least squares (linear-algebra ch.12 §9) applied to the degradation model's $H\mathbf x=\mathbf b$ .