A database of visual images includes metadata having, for a particular image, at least one entry specifying: a part of that image, another stored image, and a measure S.sub.abi of the degree of similarity between that specified part and the specified other image. One or more images are displayed and input is received from a user (for examples, by using a gaze tracker) indicative of part of the displayed images. Measures of interest are determined for each of a plurality of non-displayed stored images specified by the metadata for the displayed image(s), as a function of the similarity measure(s) and the relationship between the user input and the part specified. On the basis of these measures, a selection is made from those non-displayed stored images of further images for display.