Performance comparisons and information theoretic formulation for layered videoconferencing /
The focus of this research is on the performance of rate-constrained layered coding of video, i.e. layered video coding methods where the combined rate of the layers is subject to a fixed rate constraint, restricted to two-layer coders consisting of a baseline H.261-coded layer and a single refineme...
| Main Author: | |
|---|---|
| Format: | Thesis Book |
| Language: | English |
| Published: |
[Place of publication not identified] :
[publisher not identified] ;
1995.
|
| Subjects: | |
| Online Access: | http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=742535271&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD |
| Summary: | The focus of this research is on the performance of rate-constrained layered coding of video, i.e. layered video coding methods where the combined rate of the layers is subject to a fixed rate constraint, restricted to two-layer coders consisting of a baseline H.261-coded layer and a single refinement laye r. A novel error block weighting scheme is first introduced as a means of defining perceptually important regions of a video frame. The weighting scheme is used in conjunction with existing layered coding methods utilizing reconstruction error refinement and quantized transform coefficient refinement in order to establish a class of rate-constrained layered coding methods. We then introduce a new layered coding method which instead refines the motion vectors present in the baseline layer coder, an idea previously unexplored in the literature, where the (coarse) macroblock-based motion prediction in the baseline H.261 layer is refined at the block level in order to provide a more accurate representation of the motion present within a frame. Simulation results are presented that demonstrate the gain obtainable from this method of refinement above that obtainable from the other methods investigated. In order to obtain an idea of how well these coders perform, a reasonable estimate of the optimal obtainable performance must be found. To this end, a novel information-theoretic framework called the refined descriptions problem is introduced in order to establish performance bounds for layered coding.in general. A mathematical basis for refined descriptions is first defined, combining the classic ideas in information theory of multiple descriptions and successive refinement in a manner that naturally lends itself to a description of layered video coding in a heuristic sense. Analytical results are obtained describing the information-theoretic performance bounds for the classic i.i.d. Gaussian source in order to establish limits on the performance of "real" video sources. |
|---|---|
| Item Description: | Vita. "Major Subject: Electrical Engineering". |
| Physical Description: | xv, 125 leaves : illustrations ; 28 cm. Issued also on microfiche from University Microfilms Inc. |
| Bibliography: | Includes bibliographical references. |