| Skip to main content | Skip to sub navigation |

This is now an inactive research group it's members have moved on. You can find them at their new research groups:

ECS Intranet:

Systems for storing, retrieving, browsing and processing digital multimedia information are increasing in number rapidly as facilities for capturing digital images, digital audio and digital video become cheaper and easier to use. The development of versatile facilities for effective handling of this material offers many unsolved research opportunities and challenges. Real-time processing enables direct support of people interacting with their environment, so that images, video and audio can be used to link digital and physical worlds.

Our research involves content analysis of a wide range of multimedia including: images, video, audio and augmented reality. We aim to build systems which provide more powerful content based browsing, retrieval and navigation. More intelligent facilities are being investigated to give direct access to the semantics of the media. This involves the solution of many problems in computer vision, audio (speech and music) understanding and 3-D visualisation. Examples of current work include the development of high resolution image retrieval and navigation systems for art gallery collections, the handling of continuous metadata for continuous media streams, the development of content based retrieval of 2D/3D and the combination of images and 3D scans to characterise art works.

Combining the automatic extraction, representation and manipulation of semantic content with more traditional information handling facilities in distributed environments requires new multimedia system architectures and these are also being investigated.

The development of integrated cross-media content and concept based browsing, retrieval and navigation for distributed multimedia is continuing to provide a challenging long-range research goal.

Research Areas

Content and Concept Based Retrieval and Navigation

This program of research is concerned with media processing, extraction of semantics and architectural developments to provide more effective multimedia systems using content and concept based browsing, retrieval and navigation techniques. The development of more powerful image matching algorithms continues to play an important role here but the use of knowledge bases and creating associations between visual information and other forms of knowledge is an important aspect of the work.


We have released open source code to allow processing and web browsing of high resolution and multispectral images (VIPS and IIPImage). Past projects include imaging research is to make a "fingerprint" of art works using a combination of multispectral and 3D imaging. Recently a new PTM rig for scanning objects was built together with Archaeology.

Audio Understanding and Audio on Demand

Our goal is to incorporate audio in the distributed multimedia information space and interact with it live. Work is in progress to develop tools to extract identifying features from audio files by applying domain-specific knowledge to the analysis, in particular using the rich variety of representations for musical content. Audio is a particular focus of our presence research, in which we are studying the use of private and public audio spaces, and the use of audio for remote participation in collaborative events.

Video Databases

We are developing tools and techniques to facilitate the retrieval of information from video databases and the interpretation and presentation of that information with hypermedia

Augmented Reality

Augmented reality technologies raise many research challenges, including their integration with information systems. For example we are investigating the use of glasses which superimpose images and text onto the real world scene through the use of a camera which maps 3D graphics into the real coordinate frame. Our current research in this area aims to augment views of 3D objects with associated metadata in an intelligent and dynamic way. By looking up contextually appropriate links dynamically the user can interact with the objects in new ways.