Use the `create Tool` to generate an image with uniform random noise on each pixel.
Threshold the resulting image to obtain a binary image with 10 percent zero values.
Multiply an image by this mask image. This will generate a simulated effect of pixel drop out.

- Now apply the median filter and again assess the quantity of remaining pixel noise.
Repeat the experiment for various percentages of drop out.
- Plot a graph of the number of remaining drop out pixels as a function of the percentage
of simulated drop out. At what point does the algorithm begin to fail to remove all drop out noise?

root 2018-09-26