Added NumPy & PyTorch based denoising scripts.

[MatthewDowicz]

What?

Added functionality to denoise the FVC images. Functionality implemented by using a Neural Network architecture. It includes models, utility directory, and a notebook.

Why?

Aid the FP team in their daily FVC image inspections.

How?

This includes:

• models, written in PyTorch (for speed) and Numpy (for stability), that the user can select.
• A models utility directory that houses necessary scripts for the models to run.
• A Jupyter Notebook that runs these models on noisy data to act as a test/example of how to use these models.

Testing?

I ran tests of the scripts in the Jupyter Notebook on a stack of 70 exposures.

Anything Else?

Note that the numpy based denoiser takes about ~14 minutes to run.