Skip to main content

Architectures for Media Appliances

Advanced media appliances employ media processing technology to capture, transform and display real world information. These products will be able to carry out tasks such as: 3D capture (e.g. a portable, self contained object scanner for generating 3D models of real world objects or environments); enhanced human computer interaction and communication (e.g. virtual viewpoint video for enhanced teleconferencing, viewer tracking, face recognition, model based low bandwidth talking head transmission); differentiated media capabilities (e.g. immersive/responsive photorealistic video for education and entertainment).

Our interest is in the emerging generation of consumer computer products. All these applications would store vast amounts of media computation and consume and produce vast amounts of media data. Appliances are severely constrained in terms of cost and power. We intend to answer the question ``what can we do in an appliance, and how?''

The approach is essentially pragmatic, targeting the processing system and the application in parallel. Picking one or two applications, we will break apart the media processing into components (for example, scene analysis, warping, stitching, extraction and similar functions) and understand how they interact. We will explore scalable processing techniques that will let us build systems of different levels of functionality and performance. We will build demonstrators early to ensure that we really understand the level of processing that can be squeezed into an appliance.

Staff and Students

Mark Everingham, David May, Majid Mirmehdi, Henk Muller, Robin Gallimore(HP), Nick Wainwright(HP), Stuart Quick (HP), Stephen Pollard (HP)
Matt Patterson, Paul Clark, Dan Towner.

Collaborators

Hewlett Packard Laboratories Europe at Bristol.

Support

Hewlett Packard Laboratories Europe at Bristol.