Recent advances in the availability of 3G-cellular phones with imaging and multimedia capability and in the development of the new mobile telephone networks opens completely new horizons for the development of innovative IT tools in the domain of cultural heritage, with particular emphasis to archaeological sites and museums. AGAMEMNON is a project, co-funded under the 1ST 6th Framework Program of the European Commission, which exploits mobile phones equipped with embedded cameras as the main input/output device to provide a visitor of an archeological site with a personalized, information enriched experience. In this paper we describe the image-based "attention awareness engine" which represents the main concept entrusted into AGAMEMNON's user interface a mechanism for querying multimedia cultural heritage contents based on the user's context. We shall also discuss the main issues underlying the addressed problem, and review the preliminary results obtained so far in the project development.