samsvp / circle_painter Goto Github PK

This program uses a background image to recreate a new target image. A binary mask must also be passed to tell the program which parts to recreate.

The program selects pixels from the target image and then calculates the average color of different regions from the background image. The region with the average color closest to the selected pixel value is then placed where the pixel is on the image.

First compile the script

./compile.sh

Then run

./main -t target_path -m mask_path -b background_path -a alpha -s steps

Where

• target_path is where the image to be painted is;
• mask_path is where the mask to what parts of the image to paint is;
• background_path is the path of the image to use to reconstruct the target;
• alpha is the maximum alpha (between 0 and 1);
• step is the amount of pixels to move when taking circles from the image;

There is also a (much) slower python version (it is not a wrapper, but instead a rewrite) Run

python3 main.py -t target_path -m mask_path -b background_path -a alpha -s steps

The C++ program uses opencv and openmp. It is the recommended way to run, as it is much faster than the python implementation, specially on bigger images.

The python program uses skimage, numpy and nptyping.