Cultural heritage often fails to represent a successful attraction because it is not able to capture the interest of tourists. Sometimes the presentation of an archaeological site cannot be improved because of legal or environmental constraints. Sometimes it is really difficult to recognize the relevant details or to understand its significance. A new way of exploiting archaeological sites in a more informative and intuitive way is needed. It is necessary to provide people with information and contents, for understanding the world around them, and with the possibility to interact with those places or objects in real time. Image recognition and analysis is a promising technology that can be applied in this field. In this paper we use in indoor, but also in outdoor scenarios, image recognition for the content discovery service. In particular search by sample facilities are speed-up and improved using a position based filtering. Moreover it is used to extract semantic from subjects in order to personalize the discovery and recommendation according to the user's profile.