Colorization algorithm for grayscale image by propagating seed pixels

This paper presents a method of the colorization for a black and white imagery by propagating seed pixels. Although the colorization is an ill-posed problem, in this paper, the problem is solved based on local Markov property on images. At first, a user plants some color pixels as seeds on a grayscale image. Then, colors are propagated from the seed pixels by minimizing the color difference among connected 4-pixels. In this paper, a basic propagating algorithm is proposed and modified algorithms, which can stop the propagation at edge parts by setting partitions, are introduced. By comparing with a previous algorithm, the proposed algorithm works very well for both quality and speed.