Share Email Print

Proceedings Paper

RD-optimized competition scheme for efficient motion prediction
Author(s): J. Jung; G. Laroche; B. Pesquet
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

H.264/MPEG4-AVC is the latest video codec provided by the Joint Video Team, gathering ITU-T and ISO/IEC experts. Technically there are no drastic changes compared to its predecessors H.263 and MPEG-4 part 2. It however significantly reduces the bitrate and seems to be progressively adopted by the market. The gain mainly results from the addition of efficient motion compensation tools, variable block sizes, multiple reference frames, 1/4-pel motion accuracy and powerful Skip and Direct modes. A close study of the bits repartition in the bitstream reveals that motion information can represent up to 40% of the total bitstream. As a consequence reduction of motion cost is a priority for future enhancements. This paper proposes a competition-based scheme for the prediction of the motion. It impacts the selection of the motion vectors, based on a modified rate-distortion criterion, for the Inter modes and for the Skip mode. Combined spatial and temporal predictors take benefit of temporal redundancies, where the spatial median usually fails. An average 7% bitrate saving compared to a standard H.264/MPEG4-AVC codec is reported. In addition, on the fly adaptation of the set of predictors is proposed and preliminary results are provided.

Paper Details

Date Published: 29 January 2007
PDF: 11 pages
Proc. SPIE 6508, Visual Communications and Image Processing 2007, 650815 (29 January 2007); doi: 10.1117/12.714374
Show Author Affiliations
J. Jung, Orange-France Telecom R&D (France)
G. Laroche, Orange-France Telecom R&D (France)
École Nationale Supérieure des Télécommunications Paris (France)
B. Pesquet, École Nationale Supérieure des Télécommunications Paris (France)

Published in SPIE Proceedings Vol. 6508:
Visual Communications and Image Processing 2007
Chang Wen Chen; Dan Schonfeld; Jiebo Luo, Editor(s)

© SPIE. Terms of Use
Back to Top