TRANSFORMATION OF FILM-GRAIN NOISE INTO SIGNAL-INDEPENDENT ADDITIVE GAUSSIAN-NOISE

A point transformation is applied to images degraded by film-grain noise in order to transform the noise into approximately signal-independent additive Gaussian noise. The condition for validity of this technique is that the emulsion noise follow the square-root law predicted by the classical model of film-grain noise. It is also assumed that the scanning aperture is large compared with the grain size. Once the image has been transformed in this way, the appropriate least-squares filtering technique may be applied to the image. Experiments on two Kodak emulsions indicate that the technique works well over a range of diffuse densities up to densities of 2.