Automatic estimation of parametric signal-dependent noise models for adaptive local-statistics filtering

In this paper, a class of signal-dependent noise models that are encountered in image processing applications is considered. Such models are uniquely defined by the gamma exponent, which rules the dependence on the signal, and by the variance of a zero-mean random noise process. An automatic procedure for measuring such parameters directly from noisy images is presented. Then, LLMMSE filtering is defined and applied in a multiresolution fashion, to take advantage of increasing SNR of the data at decreasing resolution. A rational Laplacian pyramid is generalized to the noise model to yield noise that is independent of the signal on its layers. Experiments on noisy images show a high accuracy of results, both of noise estimation and filtering.