Mosaic Maker - OpenCV
About
The repository contains materials for my classes at MIM UW.
The subject of these classes is the implementation of mosaic generator. Take a look at the picture of the cat below.
What looks like a pixelated picture from the DOS era, is a composition made of a lot of the smaller cats images:
You can find the full-size image version at
notebooks/assets/0-intro/cat-output.jpg.
I used the cat dataset (available also here) to create the image above. You can find the original image here.
The aim is to create an algorithm that generates mosaic from the target and source images given as an input. By target image, I mean image that we want to recreate and by source images a set of pictures that will be used for the final composition.
About the algorithm
The algorithm presented here is by no means any competition to the state of art mosaic creation solutions. This project is meant to introduce basic Numpy and OpenCV functions in a fun, exciting way, not to build the best mosaic creation algorithm available. On the bright side, it is relatively fast thanks to histograms usage :-)
You can find a description of the algorithm is in the Jupyter notebooks (see the next section for details).
How to use this repository?
The master branch contains a basic algorithm structure without crucial parts
implemented. Instead, you will find ToDo items describing what should be
implemented in their place together with links to the documentation.
Jupyter notebooks describe step-by-step implementation instructions (you can
find them in the notebooks directory). Implementing the solution by following
notebooks is the recommended way of working with this repository.
If you are stuck with progress, you can refer to the release/solution branch
that contains a full implementation of the project.
I recommend reading the Numpy quickstart manual before starting the work.
In case of any issues or remarks, feel free to contact me.