Patch Based Wiener filter for Image Denoising

In this paper, we propose an image denoising method based on performance limits analysis for denoising of images. Peak performing real-time denoising algorithm is derived from the earlier analysis. A wiener filter is proposed based on patches which utilize redundancy in patches for denoising of images. Different parameters of the filter are estimated using the geometrical and photometrical similar patches. We illustrate that from the input noisy image we can estimate these parameters. Our denoising algorithm, intended for near-ideal performance has credible statistical base that is deliberated in detail. The efficacy of algorithm is experimentally demonstrated on different images & noise models. The results demonstrated here imply that our proposed method is unique.