Scalable Video Coding Based on High Efficiency Video Coding (HEVC)
+
Hyomin Choi,
+
Junghak Nam,
+
Donggyu Sim, and *Ivan V. Baji
+
Kwangwoon University, *Simon Fraser University
hyomin06@kw.ac.kr, qejixfyza@kw.ac.kr, dgsim@kw.ac.kr, *ibajic@ensc.sfu.ca
Abstract
In this paper, we propose two structures for
scalable video coding (SVC) based on HEVC. Several
inter-layer prediction mechanisms are introduced to
improve coding efficiency of the proposed HEVC-
based SVC. The proposed inter-layer predictions are
developed on single-loop and multi-loop decoding
structures. We found that the proposed SVC is able to
decrease average bitrates of enhancement layers by
about 10.2% for the all-intra case, and 7.4% for the
random access case, compared with single layer
coding with no inter-layer prediction in multi-loop
decoding. In addition, the single-loop decoding with
the proposed inter-layer predictions achieves coding
gains of 10.2% for the all-intra case, and 2.6% for the
random access case.
1. Introduction
After concluding H.264/AVC standardization, joint
video team (JVT) considered extension standards for
various services and niche markets such as studio,
multi-view video, and scalable applications. The
H.264/AVC scalable extension was one of the
outcomes of those efforts. The aim of scalable video
coding is to efficiently support various video qualities,
resolutions and frame rates under heterogeneous
channel conditions. These days, people carry diverse
multimedia devices and prefer to use wireless access
for improved mobility. Screen resolution is also
gradually increasing to provide higher quality for end-
point users. To respond to these changes, ISO/IEC
MPEG and ITU-T VCEG have formed joint
collaborative team on video coding (JCT-VC) with
similar purposes of JVT, and have undertaken a new
video coding standardization effort, which is called
high efficiency video coding (HEVC). Up to now, the
test model HM3.0 (reference software) and the
working draft document (WD3.0) have been released.
Note that the final draft international standard of the
HEVC will be released in January 2013. Recently,
JCT-VC is also discussing scalable video coding
(SVC) based on HEVC. In the 5
th
JCT-VC Geneva
meeting, several contributions for scalable video
extensions were presented.
H.264/AVC scalable extension employs single-loop
decoding to reduce decoder complexity. This
restriction leads to rate-distortion performance loss
compared with multi-loop decoding. However,
H.264/AVC multi-view extension adopted a multi-loop
decoding approach. The single-loop decoding design
means that all layers that are coded at the encoder are
not necessarily decoded at the decoder. Only the target
layer is mandatory to be reconstructed. On the other
hand, the multi-loop decoding design means that all
layers are reconstructed at the decoder. In this paper,
we present two scalable video coders based on HEVC
with single-loop and multi-loop decoding for
improving coding efficiency. Multi-loop decoding can
support not only scalable extension, but also multi-
view extension of HEVC, with one consolidated
coding structure.
The paper is organized as follows. First,
fundamental aspects of H.264/AVC scalable extension
are presented in Section 2. In Section 3, we present
inter-layer prediction mechanisms of the proposed
SVC based on HEVC for single-loop and multi-loop
decoding. Experimental results are presented in
Section 4, followed by conclusions in Section 5.
2. H.264/AVC Scalable Extension
Scalable video coding aims to remove the
redundancy between multiple layers of scalability:
spatial, temporal, and quality. For minimizing the
amount of redundant information between the layers,
three kinds of inter-layer prediction were adopted for
SVC based on H.264/AVC, namely inter-layer texture
prediction, inter-layer motion prediction, and inter-
layer residual prediction.
346 978-1-4577-0253-2/11/$26.00 ©2011 IEEE