Colour for image retrieval and image browsing

Content-based image retrieval, which is based on the principle of extracting image features and visual similarity based on these features, has been an active research area for the last two decades. While several kinds of features can be used for retrieval, colour features are the most widely and most successfully employed, and are hence at the heart of various image retrieval engines. However, it is also known that colour is not a stable image cue as it depends on various confounding factors including scene illumination and device characteristics. In this paper we present solutions to this problem in the form of colour invariant features, i.e. features that do not change with a change of light, putting particular emphasis on colour invariance for uncalibrated images. We then look at the use of colour not for direct retrieval, but for image database visualisation and navigation, and present some colour-based image browsing systems which allow for intuitive and efficient exploration of large image repositories.