###########################
 Image manipulation tricks
###########################

**************
 Segmentation
**************

Ostu's method
=============

.. note::

   **References**

   -  `Implement Otsu's thresholding in Python without using OpenCV and
      MATLAB
      <https://www.quora.com/How-do-I-implement-Otsus-thresholding-in-Python-without-using-OpenCV-and-MATLAB-1>`_

   -  `Scikit's skimage.filters.threshold_otsu
      <https://github.com/scikit-image/scikit-image/blob/master/skimage/filters/thresholding.py#L230>`_

   -  `Scikit's Otsu's thresholding
      <https://scipy-lectures.org/packages/scikit-image/auto_examples/plot_threshold.html>`_

:download:`source: otsu.py <otsu.py>`

.. image:: otsu.svg
   :width: 800px
   :align: center

This example uses Otsu's method to combine two sources (two different
images of the same object) to get an optimal threshold. The idea is to
get a weighted average of two images as follows

.. math::

   \mathcal{I}
   =
   (1 - \alpha) \mathcal{I}_a + \alpha \mathcal{I}_b

whereby :math:`\alpha` is chosen such that the separability, defined by
Otsu is maximised.