Automatic identification of noises for an optimal filtering

The author proposes a method which allows detection of the additive, multiplicative, or impulsive nature of noise by decision criteria from the degraded image. This method is based on the analysis of local variations of the average and of the standard deviation estimated on the homogeneous regions of the observed image. In the case where additive or multiplicative noise is detected, its statistical parameters are estimated from the local variations determined in the identification phase. This makes it possible to apply the best adapted filtering algorithms to the image to be processed