Estimation of image noise variance

A novel algorithm for estimating the noise variance of an image is presente d. The image is assumed to be corrupted by Gaussian distributed noise. The algorithm estimates the noise variance in three steps. At first the noisy i mage is filtered by a horizontal and a vertical difference operator to supp ress the influence of the (unknown) original image. In a second step a hist ogram of local signal variances is computed. Finally a statistical evaluati on of the histogram provides the desired estimation value. For a comparison with several previously published estimation methods an ensemble of 128 na tural and artificial test images is used. It is shown that with the novel a lgorithm more accurate results can be achieved.