Foveated Non-local Means Image Denoising Using Wiener filter Center Weight

Inspired by the human visual system (HVS), Alessandro Foi et al. design a foveation operator to calculate the similarity of patch. The calculation of similarity is called the foveated self-similarity (FSS). Their study shows that FSS is much superior to the traditional windowed self-similarity (WSS). Non-local means (NLM) with FSS (NLM-FSS) produces a more satisfactory denoising effect. However, as with many NLMs with WSS (NLM-WSS), center weight is not considered sufficiently. This study shows that window based Wiener filter center weight proposed by Zhang also can improve the NLM-FSS.