Community Research and Development Information Service - CORDIS

Automatic feature extraction for efficient retrieval in large multimedia database

FIR's approach to image retrieval is based on a mixed use of traditional techniques (textual classification of images) with the modern Wavelet technology, which allows image retrieval to be performed through pictorial contents. FIR reduces significantly costs and times of retrievals. FIR allows remote access to the image database through Internet, exploiting the world wide web (WWW) client-server approach and the Java support. The current FIR system is specifically addressed to electronic publishing and image trading, which treat daily huge image databases and bear high costs and inefficiency due to the lack of automation and intrinsic complexity of image handling. Nevertheless the FIR technology has a general applicability to all domains sharing the problem of image handling. The FIR technology is effective for needs of electronic publishers or anybody else supposed to handle large image databases, due to the combination of Wavelets with textual annotation. Wavelets alone concentrates only on image syntax (layout, shape, colour distribution etc), while textual annotation concentrates only on an image discrete semantic (textual attributes describing concepts represented in the image, eg feelings or city names). The FIR solution proposes a mix of both image syntax and semantics information, paying attention to their main differences. This approach allows reducing time/costs of insertions and retrievals, optimizing also retrieval results. FIR also enables a model of an image trading agency to be defined, co-ordinating offers of image providers with customers' needs, through the support of an innovative system that greatly reduces time and costs for accessing useful images. The demonstrator also provides facilities for remote access to the image database, integration also specific technologies such as WWW and Java applets. This aspect demonstrates the capability of FIR to support the building of a European Super Agency in a new image trading market place.

Reported by

Via Frigia 27
See on map
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top