An OpenCV computer vision tool for advanced image inversion.
- Add keyboard shorctut and pipe straight into application
- Add buttons for modes (e.g., invert no hue, invert, swap white/black, invert all contours).
- Add keyboard shortcut to docs
- Add crosshair cursor
- Fix slider labels
- Add rectangle indicator when making selections
- Add ctrl-z
- Fix middle-mouse click bug
- Change default behaviour
I use a system-wide dark gruvbox theme, and so I like all my applications to use a dark theme. This includes my note-taking application which compiles markdown using a custom stylesheet. This stylesheet is configured with gruvbox colours, and so notes look like this:
However, when I paste an image from somewhere on the internet, chances are it will have an ugly white background.
One solution is to add a css rule to invert every image, which results in this:
I'd like an easy way to invert only parts of an image, and that's why I created this tool, which gives the following image,
- NumPy
- OpenCV2
$ git clone https://github.com/deanpoulos/BoxInvert.git
$ cd BoxInvert
$ pip install .$ box-invert <some_image>$ box-invert BoxImage/img/img3.png
Drag the top three sliders to adjust automatic inversion thresholding, in case of a simple image.
Identify rectangular features to be inverted, click and drag a rectangle around them to invert it.
Shift+click and drag rectangular region to un-invert selection.
For features which are not rectangular, toggle on "Draw Contours" by sliding the slider to the end.
Drag the "Contours Threshold" slider until the green selection tightly fits the non-rectangular area you'd like to invert (in this case, it is the masked head in the second part of the pipeline).
Ctrl-click and drag a rectangular selection around the contoured region you would like inverted (again, it is the masked head in the second part of the pipeline). Now, toggle-off "Draw Contours".
If the final image is being placed in a notebook which has an image-inversion rule in its css, you will need to invert the image in preparation. Slide the "Invert Output" slider to the end, then right-click the image, then copy it to clipboard or save it.
