With the advances in multimedia databases and the popularization of the Int
ernet, it is now possible to access large image and video repositories dist
ributed throughout the world. One of the challenging problems in such an ac
cess is how the information in the respective databases can be summarized t
o enable an intelligent selection of relevant database sites based on visua
l queries. This paper presents an approach to solve this problem based on i
mage content-based indexing of a metadatabase at a query distribution serve
r. The metadatabase records a summary of the visual content of the images i
n each database through image templates and statistical features characteri
zing the similarity distributions of the images. The selection of the datab
ases is done by searching the metadatabase using a ranking algorithm that u
ses query similarity to a template and the features of the databases associ
ated with the template. Two selection approaches, termed mean-based and his
togram-based approaches, are presented. The database selection mechanisms h
ave been implemented in a metaserver, and extensive experiments have been p
erformed to demonstrate the effectiveness of the database selection approac
hes.