A fair ranking method for image database retrieval

This work aims at organizing the results of query-by-example image database management system so that also the less relevant objects still representing a minority, are presented to the user. Usually, during image retrieval, the objects very similar to the query are ranked in the frist positions and shown to the user. On the contrary, objects that slightly difer from the target image, and that are less numerous, are hardly ever shown to the user. The proposed method increases the chances for a fair retrieval of multimedia data from digital databases.