Master 2 SI - Mathematical MorphologyConnected operators
Problem 1 (without a computer): connected filtering a synthetic image
Let $F$ be the following image:
- Represent the $k$-section $F_k$ of $F$ with $k=0,\ldots,9$.
- Label the connected components of $F$ (letters can be used as labels), assuming that the image is equipped with the 4-adjacency.
- Draw the component tree of $F$.
- Simplify the image $F$ by iteratively razing the maxima of minimal area, until the image contains a unique maxima.
- Same question but according to depth.
- Same question but according to volume.
- Draw the unique maximum of the image obtained in the previous question.
Problem 2 - Using and designing connected filters to filter images
Download the following Python notebook or open it in Colab and follow the instruction in the notebook.
In order to open and edit the notebook on your machine, you need Python 3.5+ with Jupyter and common scientif packages (numpy, matplotlib, sklearn, imageio). If you don"t have a Python distribution installed on your machine I recommend to install Anaconda. In a terminal, with Python availbale in the PATH (eg. launch Anaconda prompt), go in the directory containing the notebook, execute the command
jupyter notebook
and open the notebook in the newly opened navigator tab.
Benjamin Perret 2020 strongly inspired from J. Cousty practical sessions .