Restoration of altered images is an ill-posed problem. Even when the blur is known, the noise leads to uncertainty. The naïve approach fails since it retains a lot of noise in high frequencies that destroys the computed restored image. Instead of using the naïve approach, we use spectral filters that weigh the components of the image that are not so contaminated by noise. In this work, we create new filters that we use with a new statistical method that was developed. For that method, we assume that the Discrete Picard condition is satisfied and we define the Picard Parameter using statistical properties of the noise. This new method gives a way to quantify uncertainty. Experiments show that our new method with our filters works comparably if not better than the already existing ones.