Scene Painting Using Photographs

CGM | Technion - Israel Institute of Technology

   Project Description

In this work, a method of colorizing a gray picture using a large pool of colored images is presented.
The goal of the project is to colorize a gray picture according to the colors of a colorized picture, which is chosen from the pool.

 

This project is a continuation and enhancement of the project “Scene Painting Using Photographs”, which was done by Kian Tarbush and Salman Bader.

 

The general problem of colorizing a gray picture has no exact solution. In most of the cases an interference of the user is required. In this project, an automatic approach is presented.

 

There are two main approaches to the problem of colorizing a grayscale picture. The first approach is based on transferring a color from a source image to the grayscale image by matching different kinds of information between the images.

The second approach is based on annotating the grayscale image with a few color scribbles and then automatically propagate the colors to produce a fully colorized image.

 

This project combines the two approaches. Moreover, the colored reference image is taken automatically from a very large pool of colorized images.