scispace - formally typeset
Search or ask a question
Proceedings ArticleDOI

Performance evaluation of two state of the art DVC codecs

TL;DR: The aim is to identify strengths and weaknesses of the two codecs that can be exploited / addressed in order to improve the achieved performance relative to H.264.
Abstract: The performance of existing DVC codecs is still lacking relative to that of H.264 and work is being carried out in order to close this gap. The authors of this paper have been and still are involved in the development of two DVC codecs respectively, the performance of which is compared herein. The aim is to identify strengths and weaknesses of the two codecs that can be exploited / addressed in order to improve the achieved performance relative to H.264.

Summary (1 min read)

Introduction

  • Lacking relative to that of H.264 and work is being carried out in order to close this gap.
  • The authors of this paper have been and still are involved in the development of two DVC codecs respectively, the performance of which is compared herein.
  • This allows efficient SI generation through block-based error concealment and potentially more accurate estimation of the local correlation noise.
  • The authors have extended their work from an early extrapolation based (P type) only to one that supports bidirectional prediction for creating the side information.
  • WZ coding is performed at the frame level and in the transform-domain, on DCT-like coefficients, using turbo codes.

C. Side Information Generation

  • The generation of the SI is equivalent to an error concealment process for missing blocks (WZ blocks) in the presence of their 4-neighbours (KEY blocks).
  • The authors employ temporal and spatial error concealment methods (TEC/SEC) controlled by a mode selection algorithm as proposed in [9] for generating the SI.
  • For TEC, each missing WZ block is divided into four sub-blocks.
  • Each sub-block is concealed using two motion vectors coming from neighboring KEY blocks – where motion estimation has previously been performed – and motion vectors generated using an external boundary matching error (EBME) process , fused via a cosine weighted overlapping step.
  • The multiple prediction blocks resulting from this process form the basis for the multi-hypothesis SI coding.

D. D. Multi-hypothesis Decoding

  • The WZ decoder uses the correlation noise to calculate bit probabilities.
  • The authors estimate the correlation noise on a block basis using the 4-neighbouring KEY blocks of each WZ block.
  • Thus the final refined version of the reconstructed WZ data (Y=YN) can exhibit better image quality.
  • The encoder first divides the input video sequence into key frames Yk and WZ frames Xk.
  • The results indicate that SPI-DVC can offer a performance advantage with high motion sequences, such as Foreman.

Did you find this useful? Give us your feedback

Content maybe subject to copyright    Report

Performance evaluation of two state of the art DVC codecs
N. Anantrasirichai, D. Agrafiotis
Dept. of Electrical & Electronic Engineering
University of Bristol
Bristol, UK
{N. Anantrasirichai}{D. Agrafiotis}@bristol.ac.uk
M. Ouaret, F. Dufaux, T. Ebrahimi
Multimedia Signal Processing Group
Ecole Polytechnique Fédérale de Lausanne (EPFL)
CH-1015 Lausanne, Switzerland
{mourad.ouaret}{frederic.dufaux}{touradj.ebrahimi}@epfl.ch
Abstract The performance of existing DVC codecs is still
lacking relative to that of H.264 and work is being carried out in
order to close this gap. The authors of this paper have been and
still are involved in the development of two DVC codecs
respectively, the performance of which is compared herein. The
aim is to identify strengths and weaknesses of the two codecs that
can be exploited / addressed in order to improve the achieved
performance relative to H.264.
Keywords-distributed video coding
I.
I
NTRODUCTION
Distributed video coding (DVC) is currently seen as a
promising approach to low power, low complexity encoding
which, in theory at least [1][2], could offer similar performance
to that of typical hybrid video codecs. A common DVC
scenario involves splitting of the video frames in two
categories, KEY frames and Wyner-Ziv (WZ) frames [3][4] ,
which then undergo conventional (usually intra) and Wyner-
Ziv coding respectively. The WZ data, which may undergo
transformation, are quantized and fed to a channel coder in a
bit-plane by bit-plane fashion. At the decoder, the decoded
KEY frames are used for creating an estimate of the WZ
frames, the side information (SI). This SI is seen as the
systematic part of the channel coder output, i.e. a noisy version
of the original WZ frame. The received parity bits are used to
correct the errors present in the SI. The number of bits spent for
the WZ data largely depends on the quality of the SI and the
accuracy of the of the noise estimation.
In this paper, we evaluate and compare the current
performance of two DVC codecs that have been developed by
the EU funded DISCOVER project (2005-2007) and the UK
EPSRC funded MEDIEVaL project (2007-2010) respectively.
The comparison will potentially reveal advantages and
characteristics of the two schemes, which could then be
exploited for the design of a more efficient DVC codec.
The MEDIEVaL project has proposed a novel spatially
interleaved DVC, referred to herein as SPI-DVC. We employ
hybrid KEY/WZ frames via spatial interleaving of blocks of
pixels. This allows efficient SI generation through block-based
error concealment and potentially more accurate estimation of
the local correlation noise. We have extended our work from an
early extrapolation based (P type) only to one that supports bi-
directional prediction for creating the side information. We
have added a more sophisticated spatiotemporal concealment
(SI prediction) method, and have introduced the use of a Gray
code and of a diversity (multi-hypothesis) scheme to produce
more reliable results [5]-[7].
The DISCOVER codec is based on Transform Domain
Wyner-Ziv (TDWZ) coding, following the architecture initially
introduced in [3] and [4]. WZ coding is performed at the frame
level and in the transform-domain, on DCT-like coefficients,
using turbo codes. The SI is obtained by Motion Compensated
Temporal Interpolation (MCTI). A feedback channel is used
for decoder-side rate control. The DISCOVER codec is
publicly available here [8].
The rest of this paper is organized as follows: Section 2
describes the SPI-DVC codec. Section 3 presents the
DISCOVER codec. Results and comparisons are shown in
Section 4 followed by conclusions in Section 5.
II. SPI-DVC
The framework of the SPI-DVC codec is illustrated in
Figure 1 and described in this section.
A. Spatio-Temporal Interleaving
The first step involves splitting of the current input frame
into KEY and WZ groups in a similar fashion to the dispersed
type of flexible macroblock ordering specified in H.264. The
interleaving block size is fixed for the entire input sequence
and can range from 16x16 to 4x4 pixels. If the GOP length is
more than 2, the KEY and WZ groups alternate relative to the
previous frame so as to avoid creating potentially annoying
regions of different subjective quality. The KEY groups of two
consecutive frames are combined to avoid any significant
performance loss, especially in the case where the complex
spatial prediction modes are used, relative to full frame intra
KEY coding. The same procedure is applied to the respective
WZ groups in order for the frame length of the input to the
turbo encoder to be adequate for good performance.
B. A Gray Code
At the WZ encoder, the quantized symbols are converted
into binary data. Subsequently, before extracting into bit-
planes, the binary codes are converted to Gray codes by
XORing binary values with their logical shift-right values. At

H.264 Intra
encoder
SI-
Generation
(TEC/SEC)
SI
Deblocking
filter
Reconstructed
frame
Input
Spatial
Interleaver
WZ
Sg
T Q
Turbo
Encoder
Buffer Reconstruction
H.264 Intra
decoder
Request
bits
Decoded
Key Sg
T
T
-1
Merge
Decoded
WZ Sg
Intraframe encoder Interframe decoder
Slepian-Wolf Coder
Buffer
Gray
code
Gray
-1
code
Map
update
Turbo
Decoder
TI-Buffer
Spatial
Interleaver
KEY
Sg
2 KEYs1 KEY
Figure 1. The SPI-DVC codec
the decoder, the decoded Gray codes are converted back to
binary values in a similar fashion. The Gray code improves the
codec’s performance at the cost of a small increase in
complexity.
C. Side Information Generation
The generation of the SI is equivalent to an error
concealment process for missing blocks (WZ blocks) in the
presence of their 4-neighbours (KEY blocks). We employ
temporal and spatial error concealment methods (TEC/SEC)
controlled by a mode selection algorithm as proposed in [9] for
generating the SI. For TEC, each missing WZ block is divided
into four sub-blocks. Each sub-block is concealed using two
motion vectors coming from neighboring KEY blocks – where
motion estimation has previously been performed – and motion
vectors generated using an external boundary matching error
(EBME) process (Figure 2), fused via a cosine weighted
overlapping step. The SEC module uses bordering KEY pixels
to conceal the WZ blocks through bilinear or directional
interpolation depending on the directional entropy of
neighboring edges. The mode selection algorithm switches
between TEC and SEC based on the levels of motion
compensated activity and spatial activity in the neighborhood
of the processed block. If forward reference frames are
available, bi-directional motion estimation is employed so that
the replacement (SI) block can additionally result from
averaging a forward and backward replacement block. The
multiple prediction blocks resulting from this process form the
basis for the multi-hypothesis SI coding.
D. D. Multi-hypothesis Decoding
The WZ decoder uses the correlation noise to calculate bit
probabilities. We estimate the correlation noise on a block basis
using the 4-neighbouring KEY blocks of each WZ block. We
model the noise as a Laplacian distribution with a specific
variance that changes from block to block. The resulting
distribution should follow closely that of the difference
between the SI block and the transmitted WZ block as it
employs actual received pixels in the vicinity of the processed
block, as opposed to frame based interpolated values. As
multiple SI data are available, multiple bit probabilities can
lead to a more precise log-likelihood ratio (LLR) [10].
Moreover, having multiple SI allows usage of the correct SI to
compensate for the errors appearing in others. We define the SI
as SI
i
, i{1,..,N}, with increasing i denoting the ranking of the
SI, from best to worst, according to the MSE of the predicted
KEY group associated with each SI group. After the decoded
bit-planes are converted back to a quantisation symbol
Q
~
, the
pixel Y is reconstructed by firstly using the best SI (SI
1
) based
on its quantisation bins
1
SI
Q
. Next, the pixels of SI
2
with
QQ
SI
~
2
=
are used to replace the clipped pixels where
QQ
SI
~
1
.
The other SI is sequentially applied to the clipped pixels as
follows.
=
=
otherwiseY
QQandQQSI
Y
p
p
SI
p
SIp
p
1
1
~
~
(1)
where Y
p
is the pixel value after the p
th
SI has been received,
p{2,..,N}. Thus the final refined version of the reconstructed
WZ data (Y=Y
N
) can exhibit better image quality.
Figure 2. The TEC process
III. DISCOVER
DVC
The DISCOVER DVC architecture is illustrated in Figure
3. A detailed description and thorough performance evaluation
is given in [11]. The encoder first divides the input video
sequence into key frames Yk and WZ frames Xk. The
frequency of the key frames is determined by the Group Of
Picture (GOP) size. Key frames Yk are encoded using
conventional H.264 Intra coding. Conversely, WZ frames Xk
are transformed using the same 4x4 DCT-like transform as in
H.264. The quantized transform coefficients are then grouped
into bands and split into bitplanes which are turbo encoded.
Moreover, for each quantized bitplane a Cyclic Redundancy
Check (CRC) is computed and transmitted to the decoder.

At the decoder side, SI is generated by MCTI of the
decoded key frames [12]. MCTI includes block-based bi-
directional motion estimation, motion interpolation and spatial
motion smoothing. A virtual channel is used to model the
correlation between the DCT coefficients of the original and SI
frames. The SI is exploited by the turbo decoder. The latter
requests WZ parity bits by means of a feedback channel until
successful decoding, which is reached when the error
probability of the decoded bitplane falls below 10
-3
and its
CRC matches the one received from the encoder.
IV. RESULTS
AND
DISCUSSION
Two QCIF test sequences (Foreman, Hall) were used for
evaluating the performance of the two DVC codecs, as well as
those of H.264 Intra, H.264 inter with zero motion and H.263+
Intra. The test conditions for the DISCOVER codec can be
found here [13].
The results indicate (Figure 4) that SPI-DVC can offer a
performance advantage with high motion sequences, such as
Foreman. This perhaps is expected given the blockwise
generation of the side information and the fact that noise
estimation exploits the locally available KEY data. With
sequences of low motion (hall) it seems that the frame based
approach does better. This disparity in performance can be
traced in a number of parameters including the added overhead
of intra coding in SPI-DVC due to spatial interleaving and the
reduced frame length of the turbo encoder. The performance of
SPI-DVC should be investigated further with regards to the
following:
The coding penalty associated with intra coding a
spatiotemporally interleaved version of the sequence, and
especially when the interleaving step is small due to the QCIF
resolution used
The noise estimation and how it performs when
motion is constant or zero
The concealment algorithm for the SI generation and
how it performs when motion is low or uniform
It is positive to note that the DVC performance in both
cases is getting closer to that of H.264 inter with zero motion.
The performance of SPI-DVC exceeds that of H.264 Intra in all
cases.
V. CONCLUSIONS
This paper has examined two DVC codecs and has
evaluated their performance relative to each other and relative
to H.264 Intra and Inter with zero motion. The evaluation
DCT Quantizer
Turbo
Encoder
Buffer
Turbo
Decoder
Reconstruction IDCT
H.264/AVC
Intra Encoder
Frame
Buffer
MCTI
H.264/AVC
Intra Decoder
DCT
Y
k
X
k
X
k
Key Frames
WZ
Frames
Feedback
Channel
Wyner-Ziv Encoder Wyner-Ziv Decoder
SI
Figure 3: DISCOVER DVC architecture
Figure 4. Rate-distortion performance.

reveals the potential advantage of SPI-DVC when motion is
high, but also indicates that there is still room for improvement
especially with more static sequences, where the DISCOVER
codec performs better.
R
EFERENCES
[1]
D. Slepian and J. K. Wolf, “Noiseless Coding of
Correlated Information Sources,” IEEE Transactions on
Information Theory, vol. 19, no. 4, July 1973.
[2]
A. D. Wyner and J. Ziv, “The Rate Distortion Function
for Source Coding with Side Information at the Decoder,”
IEEE Transactions on Information Theory, vol. 22, no. 1,
January 1976.
[3]
A. Aaron, R. Zang, and B. Girod, “Wyner-Ziv Coding of
Motion Video,” in ASILOMAR Conf. on Signals and
Systems, Nov. 2002.
[4]
C. Brites, J. Ascenso, F. Pereira, “Improving Transform
Domain Wyner-Ziv Video Coding Performance”, in Proc
ICASSP, May 14-19, 2006.
[5]
D. Agrafiotis, P. Ferré, D. R. Bull, “Hybrid key/Wyner-
Ziv frames with flexible macroblock ordering for
improved low delay distributed video coding”, VCIP
2007, 2007, California, USA.
[6]
N. Anantrasirichai, D. Agrafiotis, D. R. Bull, “A
Concealment based Approach to Distributed Video
Coding”, ICIP2008, San Diego, USA.
[7]
N. Anantrasirichai, D. Agrafiotis and D. R. Bull,
“Enhanced Spatially Interleaved DVC Using Diversity
and Selective Feedback,” ICASSP2009.
[8]
http://www.discoverdvc.org/cont_Codec.html
[9]
D. Agrafiotis, D. R. Bull, N. Canagarajah, “Enhanced
error concealment with mode selection”, IEEE
Transactions on Circuits and Systems for Video
Technology, vol. 16, no. 8, pp. 960-973, August 2006.
[10]
K.Misra, S. Karande, H. Radha, "Multi-Hypothesis
Distributed Video Coding Using LDPC Codes," Proc.
Allerton Conference on Communication, Control, And
Computing, 2005.
[11]
X. Artigas, J. Ascenso, M. Dalai, S. Klomp, D. Kubasov,
M. Ouaret, “The DISCOVER codec: Architecture,
Techniques and Evaluation”, Picture Coding Symposium
2007, Lisbon, Portugal.
[12]
J. Ascenso, C. Brites, and F. Pereira, “Improving Frame
Interpolation with Spatial Motion Smoothing for Pixel
Domain Distributed Video Coding”, EURASIP
Conference on Speech and Image Processing, Multimedia
Communications and Services, Smolenice, Slovak
Republic, July 2005.
[13]
http://www.img.lx.it.pt/~discover/test_conditions.html
Citations
More filters
Journal ArticleDOI
TL;DR: Experimental results show that the proposed scheme outperforms the recently developed DVC algorithms.
Abstract: We propose an efficient selective block encoding scheme with motion information feedback in distributed video coding (DVC). The proposed scheme estimates the spatial and temporal matching costs for each block in the side information (SI) and for the blocks with high matching costs, the motion information is provided to the encoder side to selectively encode the motion-compensated frame difference signal. Experimental results show that the proposed scheme outperforms the recently developed DVC algorithms.

17 citations

Journal ArticleDOI
30 Jan 2010
TL;DR: An efficient pixel-domain WZ (PDWZ) CODEC which effectively exploits the statistical redundancy by using the code conversion and Gray code, and then reduces the channel noise by usingthe bit interleaver is proposed.
Abstract: Recently, DVC (Distributed Video Coding) is attracting a lot of research works since this enables us to implement a light-weight video encoder by distributing the high complex tasks such as motion estimation into the decoder side. In order to improve the coding efficiency of the DVC, the existing works have been focused on the efficient generation of side information (SI) or the virtual channel modeling which can describe the statistical channel noise well. But, in order to improve the overall performance, this paper proposes a new scheme that can be implemented with simple bit operations without introducing complex operation. That is, the performance of the proposed scheme is enhanced by using the fact that the Wyner-Ziv (WZ) frame and side information are highly correlated, and by reducing the effect of virtual channel noise which tends to be clustered in some regions. For this aim, this paper proposes an efficient pixel-domain WZ (PDWZ) CODEC which effectively exploits the statistical redundancy by using the code conversion and Gray code, and then reduces the channel noise by using the bit interleaver. Through computer simulations, it is shown that the proposed scheme can improve the performance up to 0.5 dB in objective visual quality.

5 citations


Additional excerpts

  • ...유사 화소 밝기 값의 표현을 사용하기 위해 사용하였고, [15]에서는 시공간 보간된 데이터를 이용하는 PDWZ에서 약간의 복잡도를 도입하여 코덱의 성능을 개선하기 위한 방안으로서 사용하였다....

    [...]

Journal ArticleDOI
30 Sep 2010
TL;DR: A new selective block encoding scheme is proposed which provides the encoder side with the motion information for the highly distorted blocks and then allows the sender to encode the motion compensated frame difference signal and it is shown that the coding efficiency of the proposed scheme reaches almost that of the conventional inter-frame coding scheme.
Abstract: Recently, DVC (Distributed Video Coding) techniques are drawing a lot of interests as one of the future research works to achieve low complexity encoding in various applications. But, due to the limited computational complexity, the performances of DVC algorithms are inferior to those of conventional international standard video coders, which use zig-zag scan, run length code, entropy code and skipped macroblock. In this paper, in order to overcome the performance limit of the DVC system, the distortion for every block is estimated when side information is found at the decoder and then we propose a new selective block encoding scheme which provides the encoder side with the motion information for the highly distorted blocks and then allows the sender to encode the motion compensated frame difference signal. Through computer simulations, it is shown that the coding efficiency of the proposed scheme reaches almost that of the conventional inter-frame coding scheme.

1 citations

References
More filters
Journal ArticleDOI
David Slepian1, Jack K. Wolf
TL;DR: The minimum number of bits per character R_X and R_Y needed to encode these sequences so that they can be faithfully reproduced under a variety of assumptions regarding the encoders and decoders is determined.
Abstract: Correlated information sequences \cdots ,X_{-1},X_0,X_1, \cdots and \cdots,Y_{-1},Y_0,Y_1, \cdots are generated by repeated independent drawings of a pair of discrete random variables X, Y from a given bivariate distribution P_{XY} (x,y) . We determine the minimum number of bits per character R_X and R_Y needed to encode these sequences so that they can be faithfully reproduced under a variety of assumptions regarding the encoders and decoders. The results, some of which are not at all obvious, are presented as an admissible rate region \mathcal{R} in the R_X - R_Y plane. They generalize a similar and well-known result for a single information sequence, namely R_X \geq H (X) for faithful reproduction.

4,165 citations

Journal ArticleDOI
TL;DR: The quantity R \ast (d) is determined, defined as the infimum ofrates R such that communication is possible in the above setting at an average distortion level not exceeding d + \varepsilon .
Abstract: Let \{(X_{k}, Y_{k}) \}^{ \infty}_{k=1} be a sequence of independent drawings of a pair of dependent random variables X, Y . Let us say that X takes values in the finite set \cal X . It is desired to encode the sequence \{X_{k}\} in blocks of length n into a binary stream of rate R , which can in turn be decoded as a sequence \{ \hat{X}_{k} \} , where \hat{X}_{k} \in \hat{ \cal X} , the reproduction alphabet. The average distortion level is (1/n) \sum^{n}_{k=1} E[D(X_{k},\hat{X}_{k})] , where D(x,\hat{x}) \geq 0, x \in {\cal X}, \hat{x} \in \hat{ \cal X} , is a preassigned distortion measure. The special assumption made here is that the decoder has access to the side information \{Y_{k}\} . In this paper we determine the quantity R \ast (d) , defined as the infimum ofrates R such that (with \varepsilon > 0 arbitrarily small and with suitably large n )communication is possible in the above setting at an average distortion level (as defined above) not exceeding d + \varepsilon . The main result is that R \ast (d) = \inf [I(X;Z) - I(Y;Z)] , where the infimum is with respect to all auxiliary random variables Z (which take values in a finite set \cal Z ) that satisfy: i) Y,Z conditionally independent given X ; ii) there exists a function f: {\cal Y} \times {\cal Z} \rightarrow \hat{ \cal X} , such that E[D(X,f(Y,Z))] \leq d . Let R_{X | Y}(d) be the rate-distortion function which results when the encoder as well as the decoder has access to the side information \{ Y_{k} \} . In nearly all cases it is shown that when d > 0 then R \ast(d) > R_{X|Y} (d) , so that knowledge of the side information at the encoder permits transmission of the \{X_{k}\} at a given distortion level using a smaller transmission rate. This is in contrast to the situation treated by Slepian and Wolf [5] where, for arbitrarily accurate reproduction of \{X_{k}\} , i.e., d = \varepsilon for any \varepsilon >0 , knowledge of the side information at the encoder does not allow a reduction of the transmission rate.

3,288 citations


"Performance evaluation of two state..." refers background in this paper

  • ...Distributed video coding (DVC) is currently seen as a promising approach to low power, low complexity encoding which, in theory at least [1][2], could offer similar performance to that of typical hybrid video codecs....

    [...]

01 Jan 1973

1,280 citations


"Performance evaluation of two state..." refers background in this paper

  • ...Distributed video coding (DVC) is currently seen as a promising approach to low power, low complexity encoding which, in theory at least [1][2], could offer similar performance to that of typical hybrid video codecs....

    [...]

Proceedings ArticleDOI
03 Nov 2002
TL;DR: This work reports the first results on a Wyner-Ziv coding scheme for motion video that uses intraframe encoding, but interframe decoding, and suggests that an asymmetric video codec could achieve similar efficiency.
Abstract: In current interframe video compression systems, the encoder performs predictive coding to exploit the similarities of successive frames. The Wyner-Ziv theorem on source coding with side information available only at the decoder suggests that an asymmetric video codec, where individual frames are encoded separately, but decoded conditionally (given temporally adjacent frames) could achieve similar efficiency. We report the first results on a Wyner-Ziv coding scheme for motion video that uses intraframe encoding, but interframe decoding.

591 citations


"Performance evaluation of two state..." refers methods in this paper

  • ...A common DVC scenario involves splitting of the video frames in two categories, KEY frames and Wyner-Ziv (WZ) frames [3][4] , which then undergo conventional (usually intra) and WynerZiv coding respectively....

    [...]

  • ...The DISCOVER codec is based on Transform Domain Wyner-Ziv (TDWZ) coding, following the architecture initially introduced in [3] and [4]....

    [...]

Proceedings Article
01 Jan 2007
TL;DR: The codec architecture and the associated tools adopted by DISCOVER (DIStributed COding for Video sERvices), a European project which has been devoted to the advancement of Distributed Video Coding for two years are introduced.
Abstract: Distributed Video Coding is becoming more and more popular among the research community, because of its interesting theoretical contributions and because there are still many open problems waiting to be solved. This paper introduces the codec architecture and the associated tools adopted by DISCOVER (DIStributed COding for Video sERvices), a European project which has been devoted to the advancement of Distributed Video Coding for two years. Along with the general description and pointers to references with more detailed information, this paper also presents some of the results obtained with the DISCOVER codec. An extended performance analysis and the codec’s executable file are both publicly available on the project’s web site www.discoverdvc.org.

494 citations


"Performance evaluation of two state..." refers background in this paper

  • ...A detailed description and thorough performance evaluation is given in [11]....

    [...]

Frequently Asked Questions (1)
Q1. What have the authors contributed in "Performance evaluation of two state of the art dvc codecs" ?

The authors of this paper have been and still are involved in the development of two DVC codecs respectively, the performance of which is compared herein. The aim is to identify strengths and weaknesses of the two codecs that can be exploited / addressed in order to improve the achieved performance relative to H. 264.