This app attempts to process an image of a post-it covered wall, outputting the number of post-its grouped by colour. It does so using the following techniques:
- Filters the image to exclude that which does not satisfy a predefined RGB constraint (though this may change to an hsl constraint, see 'Todo').
- Uses edge detection to detect quadrilaterals of an allowed width and height.
As such, the app works best when post-it notes are not overlapping and when they are against a wall of a different colour to that of the post-its. It is also imperative that the colours of the post-its used are distinctly different colours.
This project was built using the Accord framework, and runs on Windows machines.
Using a Windows machine, clone the project locally and change into the CountPostIts.ConsoleUi's Debug directory.
$ git clone https://github.com/seblexis/count-post-its.git
$ cd count-post-its/CountPostIts.ConsoleUi/bin/Debug
Run the executable with the name of the image-file you are testing against.
$ CountPostIts.ConsoleUI.exe test3.jpg
The result is displayed as shown:
Images are saved for each colour found, highlighting the Post it notes the app has identified, within the \images folder.
To analyse your own images, place them in CountPostIts.ConsoleUI\Images
- Update program to use HSL values to discriminate instead of RGB values
- Have the executable in a dedicated folder rather than the Debug folder, and have the results images saved in a dedicated folder
- Refine the algorithm to be more robust.