Title
An enhanced fast mode decision model for spatial enhancement layers in scalable video coding
Abstract
Recently, the H.264/AVC standard has been extended to incorporate Scalable Video Coding (SVC). SVC offers the advantage of scalable (layered) coding, but has the disadvantage of a highly increased computational complexity at the encoder side when dealing with spatial scalability. To restrict the increase in required processing power, fast mode decision models for spatial enhancement layers have been proposed in literature. We propose a novel generic fast mode decision model for spatial enhancement layers for both P and B frames based on both the quantization of the enhancement layer and the correlation between the macroblock type in the enhancement layer and the co-located macroblock in the reference layer. In this paper, an evaluation of the proposed model and comparison with a state-of-the-art model is given. Results show that the proposed technique performs exceptionally well for spatial scalability. For both dyadic and non-dyadic spatial scalability, we achieve an average time saving of 75%, while only a slight bit rate increase of 2.23% and a minor PSNR decrease of 0.46 dB are measured. Compared with state of the art techniques, we further halve the complexity while having comparable rate-distortion results.
Year
DOI
Venue
2012
10.1007/s11042-010-0716-7
Multimedia Tools Appl.
Keywords
Field
DocType
H.264,Scalable video coding,Fast mode decision,Spatial scalability,Macroblock type analysis
Computer science,Real-time computing,Coding (social sciences),Artificial intelligence,Video compression picture types,Scalable Video Coding,Macroblock,Pattern recognition,Algorithm,Encoder,Quantization (signal processing),Computational complexity theory,Scalability
Journal
Volume
Issue
ISSN
58
1
1380-7501
Citations 
PageRank 
References 
4
0.42
17
Authors
6
Name
Order
Citations
PageRank
Sebastiaan Van Leuven110313.08
Glenn Van Wallendael213723.28
Koen De Wolf3799.63
Jan De Cock438849.57
Peter Lambert553867.24
Rik Van de Walle62040238.28