Image blending is the process of blending a source onto a target image with a mask such that it appears that the source is blending nicely into the target image's background. It can be implemented in a variety of ways. I have implemented Pyramid Blending and Poisson Blending.
The repo is organized in three folders:
- Code :
main.pycontains the main blending code that works for images one to nine.getMask.pycan be used to generate a mask for a given image. - Images : This folder contains all the source, target and mask images. They are numbered one to nine.
- Results : This is the folder where the generated results are stored. All the blends for images one to nine are present here.
To run the code, Python 3 and the following libraries must be installed:
- numpy
- cv2
- scipy
- matplotlib
Then simply use the following command to run with python:
python3 main.py
Examples of the blending are given below:
Pyramid Blending is not as good as Poisson Blending, but can work well in certain cases. A good example is blending the source image of an apple:
with an orange
using Pyramid Blending results in this:
Blending the following source:
with the following target:
results in the following:
