A novel coloring framework for grayscale images

Abstract

Coloring of grayscale images finds its place in many areas such as medical imaging and coloring of black and white images and videos. However, manual coloring is time consuming and ill-posed problem. In this paper we present a semi-automatic framework for coloring of grayscale images. Given a grayscale image, the new framework operates by prompting the user to specify a source color image of similar content to that of the grayscale image. Then, it employs a large set of grey level features, K-means clustering, and a new color assignment approach to transfer color to the grayscale image. Subjective and objective evaluation revealed the capability of the proposed framework in producing faithful and true color images.