Fast and reliable indexing and retrieval of image assets
The explosion of Internet technologies has caused significant effects on the production and archive of visual data. The ever-growing digital libraries have become difficult to manage since this involves different archive organisations that may need standardisation, which is a difficult and time-consuming approach. Addressing this and access needs, content-based access is considered as an effective method to link and access various archives. Within this context, this project resulted to a sophisticated application called COMPASS that offers easy and fast image indexing and retrieval. The system has been developed on the basis of an automatic description of image visual features such as colour, texture and luminance. Thereby, users may employ sample images as queries in order to obtain the most similar images stored in the database. COMPASS is a highly efficient system that offers the creation of one million images within a couple of hours. In addition, search time for a single image in such a database is limited to just a few tenths of a second. Based on the common HTTP protocol, this user-friendly system has been extensively tested, and may be employed by organisations managing large visual databases, such as television broadcasters or news agencies.
