About
|
 |
|
IMAGINATION: Image-based Navigation in Multimedia Archives IN BRIEF IMAGINATION is a Specific Targeted Research Project (STREP), which is carried out by 8 partners under the European Union's 6th Framework programme's (6FP) Information Society Technologies (IST) priority, more specifically under the IST action line "Access to and preservation of cultural and scientific resources". The project was started in May, 2006. Its duration is 36 month and the overall budget - EUR 3,8 million. OBJECTIVES The main objective of IMAGINATION is to bring digital cultural and scientific resources closer to their users, by making user interaction image-based and context-aware. Our aim is to enable image-based navigation for digital cultural and scientific resources. Our vision is a system where users receive meaningful contextual information about images and image parts, which makes images easier to understand. Moreover, users can simply navigate to new, relevant images and texts in the knowledge repository only by clicking on interesting image parts.
To achieve this goal, IMAGINATION combines and improves existing techniques. The major instrument to provide context-sensitive, relevant information is the use of semantic metadata. Semantic metadata will be (semi-)automatically generated. To achieve the highest possible quality level of automatically generated metadata, the context of the resources (images and texts) is exploited by combining text-mining, image segmentation and image recognition algorithms. We expect that this combination will cause a synergy effect and will result in high-quality semantic metadata. The generated semantic metadata is used to visualise the context of images, and thus to make it easier to comprehend them. The metadata makes it also possible to search the repository semantically and therefore to navigate to semantically related resources in the knowledge space.
The major outcome of the project will be a complete implementation of this new and intuitive approach of navigation trough images, which will include a set of technologies and tools to support the semantic annotation of images by manual, semi-automatic and automatic techniques. OUR VISION We envision a system where the user clicks through semi-automatically annotated and segmented images. Like the paradigm of the Web, were navigation is realized by interlinked text-parts, IMAGINATION will create a Image-Web, where images and parts of the images are semi-automatically interlinked and made browsable.
Graphically enhanced contextualization through images will be carried out as shown below. image copyright: European Commission’s Audiovisual Service
|
|
|
|
|
|
|
 |
|
OUR IDEA
|
When the user moves the mouse pointer over Barroso, his name will be displyed. Once the user clicks on Barroso, the IMAGINATION system:
- automatically analyses RDF metadata embedded in the picture containing instances from an ontology: Wolfgang Schüssel, Recep Tayyip Erdogan, Alejandro Toledo Manrique;
- based on this information it consults an ontology containing background information and derives time and location context;
- based on that, retrieves articles and pictures on “Barroso” in the context of Vienna, May 2006, football, Latin America (instead of giving Barroso’s general biography or other static DB entry first).
|
|
|
|
|
