Real - Time Structured Video Decoding and Display

Increasing amounts of research are being dedicated to the representation of video sequences in terms of component parts which are rendered and composited according to scripting information. Representations chosen range from two-dimensional layers all the way through full three-dimensional databases. These types of representations show great promise for compression, interactivity, and post-production flexibility and are collectively labeled "structured video" for the purposes of this thesis. This thesis implements a flexible decoder for structured video representations. The implemented decoder supports 2D, 2-1/2D (2D with z-buffers), and 3D objects as well as explicit and parametric transformations and error signals. A simple scripting language is also created for use in testing the system. Using the environment thus created, an example structured video application, contextual resizing, is implemented and presented. Thesis Supervisor: V. Michael Bove, Jr. Title: Associate Professor of Media Technology This work was supported by the Television of Tomorrow and Movies of the Future consortia. Real-Time Structured Video Decoding and Display