Content based image retrieval and information theory: A general approach

A fundamental aspect of content-based image retrieval (CBIR) is the extract ion and the representation of a visual feature that is an effective discrim inant between pairs of images. Among the many visual features that have bee n studied, the distribution of color pixels in an image is the most common visual feature studied. The standard representation of color for content-ba sed indexing in image databases is the color histogram. Vector-based distan ce functions are used to compute the similarity between two images as the d istance between points in the color histogram space. This paper proposes an alternative real valued representation of color based on the information t heoretic concept of entropy. A theoretical presentation of image entropy is accompanied by a practical description of the merits and limitations of im age entropy compared to color histograms. Specifically, the L-1 norm for co lor histograms is shown to provide an upper bound on the difference between image entropy values. Our initial results suggest that image entropy is a promising approach to image description and representation.