The Wiener filter is Mean Squared Error (MSE) filtering that incorporates both the degradation function and the statistical characteristics of noise. The underlying assumption is that the noise and image are uncorrelated. It optimizes the filter so that MSE is minimized. In this recipe, you will learn how to implement the Wiener filter using functions from scikit-image restoration module and how to apply the filter to restore a degraded image, both in a supervised and unsupervised manner.
Restoring an image with the Wiener filter
Getting ready
In this recipe, we shall use a cactus image as input and corrupt it with noise/blur. As usual, let's first import all of the required libraries using the following lines of code...