One of the main problems associated to the management of multimedia information is the difficulty of storing and classifying the electronic information so that it can be easily retrieved afterwards. There is a significant lack of standard tools which allows the interested parties to find out what multimedia assets are already in the market regarding their subjects of interest, and even less to provide them with help to index and retrieve their existing or continuously generated multimedia information.
With regards to the media copyrights issue, the situation is even worse. Although there already exist many products in the market which permit the modification and manipulation of multimedia information, there almost exist none which may relate one object to others similar to it, or to the ones it may derive from.
For many high volume end-users of multimedia information, these problems are forbidding them the full utilisation of this technology. Companies like Prensa Española, or GIUNTI are facing the fact that, although they have a very important asset on graphic multimedia information (mainly images) gathered through all their lifetime, they can not afford the task of converting all their existing archives to an electronic format and classify and organise them accordingly.
Almost all existing products for multimedia information storage rely completely on the user judgement to classify the different assets, process that is almost always done through the use of specific keywords. This process requires a very important amount of dedication and care from the people in charge of classifying the information. Most important, as the classification is produced mainly on a personal subjective manner, the obtained results are neither standard, nor reliable. Subsequent retrieval attempts to access the stored information are difficult and many times unsuccessful.
For those companies commercialising images, there is a certain need to count with tools which allow them to automatically index and retrieve images from a set included in a database. This indexing and retrieval must be based in their graphic contents and not only in associated keywords as most of the existing market products allow. Most important, these tools must include facilities to relate one image to others because of their graphic similarities and, if possible, even to detect if one image has been manipulated from an original one. This necessity is even more important for images end-users, who need to access the information they are interested in as fast as possible, without having to look through huge images catalogues.
For those companies trying to originally create new Trademarks or Logos, there is an urgent need to access the existing accepted ones and avoid the generation of new ones which resembles them. This is even true for the different National Trademark Offices. There is an urgent need to dispose of a tool to relate them automatically and provide classification, retrieval and association facilities.
The main industrial objective of this project is to develop an application which will try to address and solve the above problems. It will do so by producing an integrated tool which will perform the semiautomatic classification of multimedia information (images) based not only in the accompanying information (footnotes, IPTC, headings, etc.) or manually entered information, but in the graphic image itself (its contents). Existing techniques, like segmentation and schema generation will be used to facilitate the task of classifying, retrieval and, mainly, relating similar images from an existing set.
It is the objective of the consortium to make use of the existing techniques already tested in the field of images classification, to improve them when necessary, and to produce an integrated application which may provide an efficient help for all kind of companies or national institutions which must address the problems of generating, distributing, archiving, accessing or protecting graphic multimedia information. In order to facilitate its subsequent commercialisation, special importance will be assigned during the development to the introduction of the necessary integration procedures and interfaces to connect the developed application to the existing company applications (distribution, billing, etc.).
Funding SchemeCSC - Cost-sharing contracts