F. Shao et al.: A Novel Rate Control Technique for Asymmetric-Quality Stereoscopic Video 1823
Contributed Paper
Manuscript received 09/09/11
Current version published 12/27/11
Electronic version published 12/27/11. 0098 3063/11/$20.00 © 2011 IEEE
A Novel Rate Control Technique for Asymmetric-Quality
Stereoscopic Video
Feng Shao, Gangyi Jiang, Mei Yu, Qiaoyan Zheng and Ken Chen
Abstract — Three-dimensional (3D) video technology is
becoming increasingly popular, as it can provide stereoscopic
perception and immersive experience to end users. Some
asymmetric stereoscopic video coding methods had been
developed by utilizing binocular psycho-visual redundancy.
However, rate control problem had not been well considered
in these methods to further control the quality of stereoscopic
video. In this paper, a novel rate control method is proposed
for asymmetric-quality stereoscopic video. In order to model
the asymmetric-quality stereoscopic video, we use a fixed
threshold to quantize the binocular psycho-visual redundancy
and establish the relationship between the distortion and
quantization for the asymmetric-quality stereoscopic video.
Then, a rate control method is engineered in stereoscopic
video coding to control the rate and quality of left and right
views. Experimental results show that the proposed method
can accurately control the rate and quality of stereoscopic
video, while having a lower computational complexity
1
.
Index Terms — Stereoscopic video, rate control, asymmetric
coding, binocular psycho-visual redundancy, just noticeable
difference.
I. INTRODUCTION
3D video is capable of providing viewers with a totally new
stereoscopic video and interactive viewing experience [1]. It
will be useful for many new applications, such as 3D
television (3DTV) and free-viewpoint television (FTV), which
becomes more popular and will be largely commercialized in
the near future [2][3]. Different 3D video formats have been
designed to represent 3D scene, among which the stereoscopic
video format has been widely used due to the fact that only
color video is involved in the representation format. However,
since the data volume of stereoscopic video is proportional to
the number of views, it requires huge storage or transmission
bandwidth.
For efficient coding of stereoscopic video, there are two
factors that can be well exploited: inter-view redundancy [4]
and binocular psycho-visual redundancy [5]. Recently, there
are some research activities in asymmetric stereoscopic video
coding by exploiting the binocular psycho-visual redundancy.
On the one hand, low-resolution right view was encoded by
1
This work is supported in part by the Natural Science Foundation of China
(grant 60872094, 60902096, 61071120), the Specialized Research Fund for
the Doctoral Program of Higher Education of China (20093305120002). It is
also sponsored by K.C.Wong Magna Fund in Ningbo University.
Feng Shao, Gangyi Jiang, Mei Yu, Qiaoyan Zheng and Ken Chen serve at
the Faculty of Information Science and Engineering, Ningbo University,
Ningbo, 315211, China (e-mail: {shaofeng, jianggangyi, yumei, chenken}@
nbu.edu.cn).
spatially down-sampling or regionally adaptive filtering to
improve the coding efficiency [6][7]. On the other hand,
various factors affecting the stereoscopic perception were
investigated by subjective perceptual experiments [8][9]. In
order to model the binocular psycho-visual redundancy, we
designed a subjective perceptual experiment and derived a
just-noticeable difference (JND) threshold between left and
right views in our previous work [10], and the right view in
stereoscopic video was encoded according to this threshold.
However, how to control the quality of left and right views in
coding was ignored.
From another perspective, rate control (RC) problem had
been widely studied in video coding standards, such as
MPEG-2 TM5 [11], H.263 TMN8 [12], MPEG-4 VM8 [13]
and H.264 RC algorithm [14]. Besides, many RC algorithms
were developed by utilizing visual attention [15] or visual
sensitivity characteristics [16]. However, to the best of our
knowledge, there are very few RC works in stereoscopic
video coding. In Ref.[17], block-wise dependent bit allocation
was proposed to address the optimal bit allocation problem in
stereoscopic image coding. In Ref.[18], a RC algorithm based
on human visual system was introduced by improving the
quadric rate-distortion (R-D) model. In Ref.[19], a multi-view
RC algorithm based on the quadratic R-D model was
proposed by analyzing coding characteristics. Since these RC
algorithms do not aim at controlling the quality of left and
right views, they can not be directly applied to control the rate
and quality simultaneously in stereoscopic video coding. In
order to characterize the rate-quantization (R-Q) relationship,
linear R-Q model [20] and quadratic R-Q model [21] were
proposed. However, these R-Q models may not necessarily
apply to stereoscopic video. Therefore, one important issue is
how to characterize the distortion-quantization (D-Q)
relationship of left and right views in stereoscopic video
coding.
For typical consumer electronic applications, such as
mobile 3DTV or 3D phone [22], there are significant
bandwidth limitations for new 3D services when an existing
HD video service is delivered [23]. For the mobile Terrestrial
Digital Multimedia Broadcasting (T-DMB) broadcasting
standard, the maximum transmission bitrate is 1.152 Mbps per
1.5 MHz of channel bandwidth [24]. The allocated rate for
one 2D video service is 512 to 544 kbps. Therefore, in order
to achieve 3D service in T-DMB, asymmetric stereoscopic
video coding is an effective way to overcome the limitations
of channel capacity. Down-sampling-based codecs have been
implemented in the developed T-DMB system. Also, the JND-
based asymmetric stereoscopic video coding method [10] can