Adaptive noise detection for image restoration with a multiple window configuration

In this paper we present a robust and efficient switch-based filtering technique that makes use of a multiple window noise detection scheme. The proposed method performs exceptionally well for both impulse and Gaussian type noise, with a minimum amount of computational expense incurred. All parameters required for the MWC detection are trained off-line using a maximum-likelihood estimator, so no a prior knowledge of the noise type or image characteristics are required. As the results show, the new technique demonstrates a marked performance gain over existing state of the art methods.