In many imaging systems the output is a noisy version of the required image due to inaccuracies in the setup or low-quality equipment. Obtaining the clean high-resolution image from its noisy version is a problem that many image processing setups have tried to deal with, some more successfully and others less. In this work the authors suggest using a large data set of images for which we know the original undamaged image, build an optimal reconstruction filter for each one using iterative methods and use this data base of filters to generate a fuzzy logic inference engine that will welcome a new unknown noisy image and by comparing it to the database generate a single reconstruction filter to denoise that image. The authors present the process of building the database and the final denoising filter and show excellent correlation results in terms of image reconstruction.