New Centre/Surround Retinex-like Method for Low-Count Poisson Image Denoising
Viacheslav E. Antsiperov, Elena R. Pavlyukova · 2023
The problem of synthesizing weak image enhancement methods is under consideration. The proposed approach to the synthesis of enhancement methods for image processing is based on the principles and methods of statistical learning, considered as learning on the input data (adaptive filtering). The synthesis is focused on a special representation of images using controlled size sample of counts (sampling representation). Based on the specifics of this representation, a generative model of an ideal image is formulated, which is then concretized to a probabilistic parametric model in the form of a system of receptive fields (Retinex model). It is demonstrated how the problem of denoising with image detail preserving could be reduced to recurrent computational procedures of the maximum likelihood approach, popular in the field of machine learning.