Image manipulation tricks

Segmentation

Ostu’s method

Note

References

• Implement Otsu’s thresholding in Python without using OpenCV and MATLAB

• Scikit’s skimage.filters.threshold_otsu

• Scikit’s Otsu’s thresholding

source: otsu.py

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

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

whereby \(\alpha\) is chosen such that the separability, defined by Otsu is maximised.