An interactive system for content-based retrieval in a pictorial database u
sing color distribution features is discussed. Images are internally repres
ented and indexed by a set of histograms reflecting homogeneous chromatic p
roperties of image structure and also including color-induced geometric fea
tures. User query images are interactively composed through a graphic envir
onment featuring color sketching, image sampling and relevance feedback. Qu
eries are first translated into the internal representation and then matche
d against the database index using a modified multi-feature histogram inter
section operator. Examples of system operation are provided, and experiment
al results are reported indicating that using multiple distributions instea
d of global color histogramming allows for a system output in better accord
ance to human expectation.