NEW: Add stainaugment feature for stain augmentation

[mostafajahanifar]

Introducing stainaugment tool for generating stain-augmented images from an input image. This class can be used stand-alone or in albumentations augmentation pipelines.

The process works as bellow:

1. H and E parts of the input images are first extracted, using a predefined stain_matrix or a stain_matrix that has been calculated based on 'Vahadane' or 'Macenko' methods.
2. Input image's stain concentrations are calculated based on the provided/extracted stain_matrix.
3. Stain concentrations are adjusted randomly based on two input sigma1 and sigma2 parameters:
   • Scaled by alpha = np.random.uniform(1 - self.sigma1, 1 + self.sigma1)
   • Biased by beta = np.random.uniform(-self.sigma2, self.sigma2)
4. New image is formed based on the adjusted concentrations.

Providing the pre-extracted stain_matrix on class initialization have two benefits:

• stainaugment can be used for small patches on-the-fly during the model training (because the bulk of stain matrix extraction or dictionary learning is eliminated)

• By setting the parameters, this function allow the user to augment the stain around a target stain matrix i.e., normalize the stain to a target image and then augment it.

• Class definition and structure

• Check the functionality

• Add support for albumentations

• Complete the docstring and writing examples

• Complete the test and coverage

• Add an example to "Stain Normalization" notebook

[codecov]

Codecov Report

Merging #181 (83295a9) into develop (b045ba4) will increase coverage by 0.00%.
The diff coverage is 100.00%.

❗ Current head 83295a9 differs from pull request most recent head 26dcf3f. Consider uploading reports for the commit 26dcf3f to get more accurate results

@@           Coverage Diff            @@
##           develop     #181   +/-   ##
========================================
  Coverage    99.79%   99.80%           
========================================
  Files           40       41    +1     
  Lines         2979     3057   +78     
  Branches       499      507    +8     
========================================
+ Hits          2973     3051   +78     
  Misses           1        1           
  Partials         5        5           

Impacted Files	Coverage Δ
tiatoolbox/tools/__init__.py	100.00% <100.00%> (ø)
tiatoolbox/tools/stainaugment.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b045ba4...26dcf3f. Read the comment docs.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[simongraham]

Nice work. Few small things to change but generally it’s nice. Happy to merge when those things are addressed

[mostafajahanifar]

Many thanks to @simongraham and @Srijay-lab for reviewing this PR. I have incorporated your comments in ff0e65c. Please let us know if you are happy with the changes (and test the code on your system if possible) so @shaneahmed can merge this PR.

[shaneahmed]

Thanks @mostafajahanifar
Please address my comments and add stain augmentation feature to usage.rst

[shaneahmed]

Thanks @mostafajahanifar