Chroma lab colourisation of grayscale images and videos

Abstract

Colourisation is a computer-aided task of adding colours to a grayscale image or video. Despite many years of research in the field of image processing, no existing standard technique can reliably perform this operation. Using present software, colourisation is a slow, intricate and expensive task, owing to the huge amount of human effort it re quires. In video colourisation an additional task is required for tracking moving features across a scene. This presents a further problem as existing solutions are not accurate and require extensive manual correction. In this final year project, we present Chroma Lab, a software solution based on latest colourisation research. This application allows the user to colourise monochromatic images and videos with minimal intervention. Chroma Lab colourises images and videos semiautomatically in a relatively short time. The user specifies colour hints which guide the colourisation process in producing sharp and realistic images. We use two different techniques. The first technique minimises the difference between the chrominance of a pixel and the weighted aver age of the chrominance of its adjacent pixels. The second technique generates the missing colours by the weighted blending of the colour hints provided by the user. Moreover, we introduce our novel idea of interpolating colour hints across frames for quick video colourisation. Using Chroma Lab, colourisation is a surprisingly trivial task, even to the common end-user. As a result, this project has the right potential to be developed in a much requested commercial package.