Stylistic indoor colour design via Bayesian network

Abstract

Colour plays an important role in interior design. A good colour scheme can usually convey some particular design philosophy and define a specific household style. Given the planar shape and functional type of a room, and a set of furniture in which a subset of furniture have been specified colours, this paper proposes a data-driven approach to automatically assign colours to the rest of the room and furniture items such that the whole colour tone is harmonious. We first train a Bayesian network to inherently encode the dependency between decorative styles and furniture colours, as well as the relevancy between furniture colours themselves, using a real interior design dataset. An optimal colourization is then obtained by maximizing the conditional joint probability of the colour assignment. This process encourages frequently used colours and punishes any deviation from the specified colours. To increase the diversity of colours, we design a strategy to jitter colours assigned to furniture items. In addition, a colour transfer scheme is adopted to support the mapping of arbitrary textures while sustaining the whole colour tone. A series of experiments demonstrate that our approach is effective and able to generate practical results.