Wiener Filtering by the Radial Power Density Function for Circularly Symmetric Blurs

It is well known that the Wiener filter is the L²-optimal linear filter for image reconstruction. It is not possible to use it in practical situations, as its exact construction requires a priori information. In this article we propose a simple approximation to the Wiener filter valid for circularly symmetric blurs which circumvents this requirement. The performance of this approximation is shown to be close to the exact case and better than the optimally constructed (in the L² sense) Tikhonov filter