intra frame coding for advanced video coding standard to … · 2017. 4. 7. · intra frame coding...
TRANSCRIPT
-
@IJMTER-2015, All rights Reserved 456
Intra Frame Coding for Advanced Video Coding Standard to reduce
Bitrate and obtain consistent PSNR Using Gaussian Pulse
Manjanaik.N1, Dr.Manjunath.R
2
1Electronics and Communication Engineering, Jain University, Bangalore, India.
2Senior Domain Specialist, Philips Company, Bangalore, India.
Abstract - This paper proposes Intra frame coding in Advanced Video Coding Standard to reduce bit
rate and achieve consistent PSNR using Gaussian pulse .In Intra prediction there are eight intra
prediction modes such as Mode-0 (vertical), Mode-1 (horizontal), Mode-2 (DC), Mode-3 (diagonal
down left), Mode-4 (diagonal doen right), Mode-5 (vertical right), Mode-6 (horizontal right), Mode-
7(vertical left) and Mode-8(horizontal up). Gaussian pulse is pulase that has waveform described by
Gaussian distribution. Gaussian pulse, which removes ringing, blocking artifacts and provides smooth
reconstructed image. The Gaussian pulse operation smoothens the signal. The proposed algorithm intra
frame coding implemented using matlab. In the proposed algorithm Gaussian pulse is multiplied with
each 4x4 transformed coefficients of intra frame of 4x4 sub macro block before quantization block.
The simulation results are obtained for yuv video sequences in CIF and QCIF format for different
quantization parameters with Gaussian values using Matlab. The proposed algorithm gives consistent
PSNR, and reduction in bitrate. The PSNR and bit rate achived for yuv intra frame sequences, are
tablated for different quantization parameters with Gaussian value. The
rate distortion curve, and bit rate vs quantization parameter are plotted. The simulation results shows
that the proposed algorithm can achieve controlled/consistent PSNR, and reduced in bit rate 24.4% for
CIF and 18.08% QCIF sequences.
Keywords: H.264, Gaussian pulse, Intraframe, PSNR, CAVLC, Prediction Modes.
I. INTRODUCTION H.264/AVC is the latest video coding standard jointly developed by Joint video team in 2003,which is
organized by two international standards bodies ie the Internatonal Telecommunication Union-
Telecommunications sector (ITU-T) video coding experts group and International Organization for
Standardization/ International Electro-technical Comission (ISO/IEC) moving picture experts
group.This standard consists of various adavanced features Intra prediction unit, integer transform,
variable block mation estimation, entropy encoding ,deblocking filter and coding tools.due to these
features this standard achieves greater compression without sacrificing on video quality. Intra frame
encoded by integer transform, Gaussian pulse, quantization and context adaptive variable length coding
to get compression ratio and bit rate. In reverse process, reconstruct the picture to measure PSNR. The
existing video compression standards such as MPEG-2, MPEG-4 support adaptive changes in the
compression ratio and bitrate. The quantisation steps are varied by the appropriate choice of quantisation
matrix coefficients. The matrix may be chosen from a set of matrices defined by the standard or user can
drive altogether a new matrix in which case the coefficients need to be transmitted along with the bit
stream. A coarse step reduces the bitrate and increases compression at the cost of drastic reduction in the
PSNR of the reproduced image. so that proposed algorithm is implemented to avoid image getting bad
hit and to achive controlled PSNR and bit rate with high compression ratio in a contolled way by
multiplying Gaussian pulse each 4x4 transformed coefficients of intra frame.
-
International Journal of Modern Trends in Engineering and Research (IJMTER) Volume 02, Issue 06, [July– 2015]ISSN (Online):2349–9745 ; ISSN (Print):2393-8161
@IJMTER-2015, All rights Reserved 457
II. METHODOLOGY
The test yuv video sequence in QCIF and CIF format is used as an input which is open source
downloaded from the website. Read intra frames of yuv video sequence frames (Foreman QCIF anc
CIF, Suzie QCIF and tempete CIF etc). Each intra frame is processed in macroblock which is further
divided into terms of 4x4 sub blocks. Each intra frame processing is done in order to get high video
quality and high compression ratio and low bit rate as compared to previous video coding standards.
Each intra frame processing includes , 4x4 sub blocks from 16x16 macro block of intra frame, residual,
integer transformation, Gaussian pulse quantization and entropy encoding at forward path. The intra
frame is reconstructed by doing inverse process at the reconstruction path.
In the proposed algorithm each 4x4 sub macro block of intra frame is transformed into frequency
domain samples using integer transformer, these samples are multiplied with Gaussian pulse, which
smoothens the signal in reversible way to avoids picture image getting bad hit, blocking artifacts and
improves functionality of quantize block, quantized coefficients are encoded using context adaptive
variable length coding to obtain compressed bit to get compression ratio and bit rate. In the reverse
process quantized coefficients are inverse quantized and inverse transformed to reconstruct the frame,
after measuring PSNR of frame.
III. BLOCK DIAGRAM The block diagram of H.264 encoder for selection of Intra prediction modes is shown in fig.1. The h.264
encoder block consists of integer transform, Gaussian pulse, quantization context adaptive variable
length coding, inverse quantization, invese transformation, intra prediction unit and deblocking filter.
Fig.1 Block diagram of H.264 encoder
The block diagram of H.264/AVC encoder includes two dataflow paths, a forward path and a
reconstruction path. An input frame is given for encoding. Every frame is processed in terms of a
Macroblock (MB) of size 16x16 pixels. Each macroblock is further sub divided into 4x4 sub
macroblock. Each 4x4 sub macroblock is encoded in intra prediction modes. A prediction macroblock P
is formed based on a reconstructed block. In intra mode, P is formed from samples in the current block
is based on previously reconstructed block. The prediction P is subtracted from the current macroblock
to produce a residual or difference macroblock. This is transformed using integer transform, transformed
-
International Journal of Modern Trends in Engineering and Research (IJMTER) Volume 02, Issue 06, [July– 2015]ISSN (Online):2349–9745 ; ISSN (Print):2393-8161
@IJMTER-2015, All rights Reserved 458
coeffients are multiplied with Gaussian pulse and quantized using quantization block to give quantized
transform coefficients. These coefficients are reordered and entropy encoded using context adaptive
variable length coding (CAVLC) and the compressed bit stream is transmitted over a band-limited serial
transmission channel . In the reconstruction path the quantized macroblock coefficients are decoded to
reconstruct a frame for encoding of other macroblocks. The quantized coefficients are inverse quantized
and inverse transformed to produce a difference macroblock. The prediction macroblock P is added to
difference maroblock to create a reconstructed macroblock after a de-blocking filter, which improves the
quality of the reconstructed frame[1-2].
IV. GAUSSIAN PULSE A pulse has the waveform of a Gaussian distribution, which is similar to a bell curve shape. Gaussian
pulse is shaped as a Gaussian function or Gaussian distribution.Gaussian pulse is based on Gaussian
function that is symmetric bell curve shape that quickly falls off towards zero.In the proposed method
the Gaussian pulse equation used for Intra frame coding in advanced video coding standard
G (v) =exp (-v ^ 2) ------(1)
The above equation are used for intra frame coding. After transform, frequency domain samples are
obtained, these samples are multiplied with a Gaussian pulse. The Gaussian operation smoothens the
samples. In the proposed algorithm the Gaussian pulses generated for different Gaussian scaling
parameters are shown in Figure.2-3. Matlab program written for generation of Gaussian pulses for
different Gaussian value, the generation of Gaussian pulse is considering the sampling frequency, time
base and Gaussian scaling parameter.
Fig.2. Gaussian pulses for v = 0.1, 0.2 and 0.4
Fig.3. Gaussian pulses for v = 1, 2 and 3
-
International Journal of Modern Trends in Engineering and Research (IJMTER) Volume 02, Issue 06, [July– 2015]ISSN (Online):2349–9745 ; ISSN (Print):2393-8161
@IJMTER-2015, All rights Reserved 459
V.INTRA FRAME CODING
The H.264/AVC intra prediction unit achieves higher compression ratio and image quality compared
with preivious standard(JPEG2000).The H.264 support different block sizes, it supports 4x4 and 16x16
block sizes for base line, main and extended profiles and 8x8 block size for high profile. There are nine
prediction modes for 4x4 blocks, four for 16x16 blocks and and two for 8x8 blocks. All the prediction
pixels are calculated based on the the reconstructed pixels of previously encoded neighbouring blocks.
The prediction of 4x4 blocks is predicted based on the previously reconstructed pixels labelled (A-M)
shown in Fig.4 the pixels (A-M) are reconstructed previously and consider as reference pixels for
current block. The pixels labeled (a-m) are prediction pixels.
Fg. 4. labeling of 4x4 prediction samples
Each 4x4 sub macroblock is predicted using eight directional prediction modes and one DC mode.The
directional prediction modes are vertical, horizontal, diagonal down left, horizontal down, diagonal
down right, vertical left, horizontal up, vertical right.For directional modes the predicted samples are
formed from a weighted average of the perdiction samples A-M.For DC mode the predicted samples are
formed by mean of samples A-D and I-L.The encoder select prediction mode for each 4x4 sub
macroblock.The selection of best prediction mode is obtained by minimizing the residual encoded block
and its prediction[3-5]. The Fig.5 shows the intra prediction modes.
Fig.5. 4x4 intra prediction modes
In H.264/AVC standard, spatial redundancy is removed in intra frame by using nine intra prediction
modes. All nine modes are used for encoding intra 4X4 luma block of intra frame, modes 0, 1, 3, 4, 5, 6,
7, 8 are directional prediction modes and mode 2 is the DC prediction mode with no direction. Mode 1
specifies the vertical prediction mode,for mode 0 prediction samples formed by subtracting
reconstructed samples (A B C D) with current block samples. Mode 1 specifies the horizontal prediction
mode, for mode 1 prediction samples formed by subtracting reconstructed samples (I J K L) with current
block samples. Mode 1 specifies DC mode prediction for 4X4 block is to replace all pixels/samples in
the current 4X4 block by the mean value of the reconstructed samples (A B C D I J K L). for modes 3 to
8 prediction samples formed from a weighted average of the prediction samples A-M.
M A B C D E F G H
I a b c d
J e f g h
K i j k l
L m n o p
-
International Journal of Modern Trends in Engineering and Research (IJMTER) Volume 02, Issue 06, [July– 2015]ISSN (Online):2349–9745 ; ISSN (Print):2393-8161
@IJMTER-2015, All rights Reserved 460
VI. OPERATION OF DC PREDICTION MODE
There are nine prediction modes in intra prediction for Intra frame coding in which one of prediction
mode steps involved are explained.the steps performed in DC mode for Intra frame coding as follows.
For DC prediction mode, all the pixels of current sub block is predicted by mean value of upper and left
reconstructed pixels of reconstructed block. Prediction equation generated for DC mode as follows
(A+B+C+D+I+J+K+L)/8 = S ----(2)
a1= a2 = a3 = a4 = a5 = a6 = a7 = a8 = a9 = a10 = a11 = a12 = a13 = a14 = a15 = a16 = S
The operation involved in DC prediction mode for intra frame is as shown in Figure 6.
Fig 6. DC intra prediction mode for intra frame
VII. STEPS OF IMPLEMENTATION
The following steps are to required for intra frame coding in AVC using Gaussian pulse.
• Input test yuv video sequence in QCIF and CIF format is used as an input .
• Read Intra frame from input test yuv video sequence (I-frame).
• Each frame is processed in terms of macro block; macro block consists of 16x16 pixels.
• A 16x16 macro block is further divided into a 4x4 sub macro block.
• A first 4x4 sub block is processed directly without using previously reconstructed block followed by integer transform, Gaussian pulse, quantization, entropy encoding at encoder and reverse process
at reconsrction path of encoder.
• Reconstruct a 4x4 sub block using inverse process with DC intra prediction mode
• Obtain residual block by subtracting current sub block with previously reconstructed sub block.
• Residual of 4x4 sub block is integer transformed, Gaussian pulse, quantized and entropy encoded at encoder and reverse process such as inverse quantized, inverse transform and add prediction at
reconstruction path of encoder.
• Finally measure PSNR, bit rate and compression ratio for different QP with Gaussian value.
-
International Journal of Modern Trends in Engineering and Research (IJMTER) Volume 02, Issue 06, [July– 2015]ISSN (Online):2349–9745 ; ISSN (Print):2393-8161
@IJMTER-2015, All rights Reserved 461
VIII. IMPLEMENTAION
The proposed work is carried out using Matlab. The input is test yuv video sequences in QCIF and CIF
format. A Matlab program is written, which reads the test yuv video sequence and extracts only Intra
frame. The next process involves intra frame is divided Macro block into 4x4sub block which is
processed directly by following usual procedure of forward path of H.264 encoder and reconstruct the
processed block which serves as reference to the next sub block using basic reverse process in
reconstruction path of H.264 encoder. Obtain the residual block by subtracting reconstructed bock with
current block. Residual coefficients are integer transformed, transformed coefficients (i.e. frequency
samples) are multiplied by Gaussian pulse, these coefficients are, quantized and encoded using context
adaptive variable length coder (CAVLC) to get compressed bit. At H.264 encoder in the reconstruction
path, perform reverse processes to get reconstruct image and finally measure quality picture (PSNR) for
Mode-2 DC prediction mode.
Measure PSNR, compression ratio and bit rate for different quantization parameters with Gaussian
value.
IX. RESULTS AND DISCUSSION
The proposed work Intra frame coding is done using Gaussian pulse. The results obtained for the yuv
video sequences in CIF and QCIF frames and different quantization parameters and Gaussian value for
Mode-2 4x4 intra DC prediction mode.
Table I Results of Foreman QCIF Hung et al [2008] algorithm
QP
PSNR (dB) Bitrate (kbits/s)
JM 8.6
(H.264)
Proposed JM 8.6
(H.264)
Proposed
24 39.92 39.95 883.72 862.48
28 37.12 37.12 619.30 598.61
32 34.30 34.29 413.40 398.21
36 31.60 31.58 274.70 263.09
Remarks PSNR 0.01 ∆Bitrate = -3.42 %
Table II Results of Foreman QCIF Proposed algorithm
QP
PSNR (dB) Bitrate (kbits/s)
JM18.
6
(H.26
4)
Proposed JM18.6
(H.264
)
Propos
ed
24 37.00 37.58 424.32 254.55
28 36.82 37.63 312.96 194.86
32 34.07 37.63 218.40 147.69
36 31.89 37.72 148.32 133.34
Remark
s
PSNR consistent ∆Bitrate = -30.04
%
-
International Journal of Modern Trends in Engineering and Research (IJMTER) Volume 02, Issue 06, [July– 2015]ISSN (Online):2349–9745 ; ISSN (Print):2393-8161
@IJMTER-2015, All rights Reserved 462
Table III Results of Suzie QCIF Hung et al [2008] algorithm
QP PSNR (dB) Bitrate (kbits/s)
JM 8.6
(H.264)
Proposed JM 8.6
(H.264)
Proposed
24 40.59 40.62 700.97 691.51
28 37.80 37.79 464.08 457.59
32 35.20 35.21 288.23 282.11
36 32.94 32.93 176.79 173.70
Remarks PSNR 0.02 ∆Bitrate = -2.11 %
Table IV Results of Suzie QCIF Proposed algorithm
QP PSNR (dB) Bitrate (kbits/s)
JM18.6
(H.264)
Proposed JM18.6
(H.264)
Proposed
24 39.85 37.31 306.16 240.24
28 37.28 37.35 208.48 180.84
32 34.94 37.38 134.36 131.75
36 32.79 37.50 86.24 96.80
Remarks PSNR consistent ∆Bitrate = -6.12 %
Table V Results of Foreman CIF Hung et al [2008] algorithm
QP PSNR (dB) Bitrate (kbits/s)
JM 8.6
(H.264)
Proposed JM 8.6
(H.264)
Proposed
24 40.25 40.27 2699.38 2659.59
28 37.77 37.76 1777.41 1746.19
32 35.18 35.18 1121.13 1102.18
36 32.87 32.86 703.62 694.30
Remarks PSNR 0.01 ∆Bitrate = -1.75 %
Table VI Results of Foreman CIF Proposed algorithm
QP PSNR (dB) Bitrate (kbits/s)
JM
18.6
(H.264)
Proposed JM18.6
(H.264)
Proposed
24 37.00 37.58 424.32 254.55
28 36.82 37.63 312.96 194.86
32 34.07 37.63 218.40 147.69
36 31.89 37.72 148.32 133.34
Remarks PSNR consistent ∆Bitrate = -
18.62%
-
International Journal of Modern Trends in Engineering and Research (IJMTER) Volume 02, Issue 06, [July– 2015]ISSN (Online):2349–9745 ; ISSN (Print):2393-8161
@IJMTER-2015, All rights Reserved 463
Table VII Results of Tempete CIF Hung et al [2008] algorithm
QP PSNR (dB) Bitrate (kbits/s)
JM 8.6
(H.264)
Propos
ed
JM
8.6
(H.26
4)
Propose
d
24 39.46 39.49 5547.
19
5502.9
3
28 36.11 36.12 4116.
46
4087.8
0
32 32.73 32.74 2843.
07
2811.5
8
36 29.63 29.63 1849.
85
1835.5
6
Remar
ks
PSNR 0.02 ∆Bitrate = -1.24
% Table VIII Results of Tempete CIF Proposed algorithm
QP PSNR (dB) Bitrate (kbits/s)
JM18.6
(H.264)
Propos
ed
JM18.6
(H.264)
Propose
d
24 38.67 35.35 2370.3
2
1363.0
5
28 35.32 35.37 1743.7
6
1104.7
2
32 31.94 35.41 1217.2
8
877.35
36 28.98 35.48 805.20 694.70
Remar
ks
PSNR consistent ∆Bitrate = -
18.62%
The results obtained for the test yuv sequences such as foreman, tempete and suzie in CIF and QCIF
format, for quantization parameter (24,28,32 and 36) with gaussian value(0.1).The table I, II,
III,IV,V,VI,VII and VIII gives PSNR, and bit rate of intra frame of yuv CIF frames for 4x4 intra DC
prediction mode with different quantization parameters and Gaussian Value 0.1. The rate distortion
curve, bitrate curve and compression ratio curve for different quantization parameters with scaling factor
v=0.1 of DC prediction mode is shown in figure 7-10.
-
International Journal of Modern Trends in Engineering and Research (IJMTER) Volume 02, Issue 06, [July– 2015]ISSN (Online):2349–9745 ; ISSN (Print):2393-8161
@IJMTER-2015, All rights Reserved 464
Table IX Performance comparison proposed method with previous work Sequence
QCIF
Method Parameters QP Bit rate reduction
Proposed
Bit rate
Reduction
Hung et al [2008]
Inference
24 28 30 36
Foreman
QCIF
JM 18.6 PSNR (dB) 37.00 36.82 34.07 31.89
-30.04 %
-3.42 %
Bit rate reduction
achieved more
Compared
With
Hung et al [2008]
PSNR is consistent
Bit rate (kbps) 424.32 312.96 218.40 218.40
Proposed PSNR (dB) 37.58 37.63 37.63 37.72
Bit rate (kbps) 254.55 194.86 147.69 133.34
Suzie
QCIF
JM 18.6 PSNR (dB) 39.85 37.28 34.94 32.79
-6.12 %
-2.11 % Bit rate (kbps) 306.16 208.48 134.36 86.24
Proposed PSNR (dB) 37.31 37.35 37.38 37.50
Bit rate (kbps) 240.24 180.84 131.75 96.80
Average JM 18.6, Intra frame, frame rate 30 frames per second CAVLC
-18.08 % -2.765%
Foreman
CIF
JM 18.6 PSNR (dB) 39.88 37.46 35.04 32.69
-18.62%
-1.75 % Bit rate (kbps) 1256.08 870.64 585.60 389.44
Proposed PSNR (dB) 38.98 38.98 39.03 39.15
Bit rate (kbps) 914.77 673.72 487.64 358.30
Tempete CIF JM 18.6 PSNR (dB) 38.67 35.32 31.94 28.98
-30.18%
-1.24 % Bit rate (kbps) 2370.32 1743.76 1217.28 805.20
Proposed PSNR (dB) 35.35 35.37 35.41 35.58
Bit rate (kbps) 1363.05 1104.72 877.35 694.70
Average JM 18.6, Intra frame, frame rate 30 frames per second CAVLC
- sign indicates saving
+ sign indicates increasing
-24.4 %
-1.49%
The graph of rate distortion for different quantization parameters with scaling factor v = 0.1 indicates
that for different quantization,consistent PSNR and reduced in bit rate achieved. The bit rate for
different quantization parameters Gaussian value v = 0.1 is indicates that for different quantization,
faster reduced in bit rate achieved compared with JM reference 18.6 software of H.264 standard.
Figure.11-12 indicates reconstructed frames Foreman and tempete in CIF format for different
quantization paremeter with Gaussian scaling factor v=0.1.
The performance parameters are measured using the following formula.
∆P = PSNR(Proposed) – PSNR(JM reference) ---(3)
∆B = Bit rate(proposed)- Bit rate(JM reference) X 100 ---(4)
Bit rate(JM reference)
Fig.7. Bitrate vs Quantization parameter
20 25 30 35 400
50
100
150
200
250
300
350
400
450
500
550Bitrate vs QP Foreman Qcif
Bitra
te (
kbps
)
Quantization Parameter (QP)
Bitrate Foreman Qcif Proposed
Bitrate Foreman Qcif JM 18.6
-
International Journal of Modern Trends in Engineering and Research (IJMTER) Volume 02, Issue 06, [July– 2015]ISSN (Online):2349–9745 ; ISSN (Print):2393-8161
@IJMTER-2015, All rights Reserved 465
Fig.8. Bitrate vs Quantization parameter
Fig.9. Bitrate vs Quantization parameter
Fig.10. Bitrate vs Quantization parameter
Table I,III, V and VII indicates the performance parameters of Hung et al [2008] algorithm.
Table II,IV, VI and VIII indicates the performance parameters of proposed algorithm.
The table IX gives over all performance comparision of proposed method and previous method.From the
table it observe that bit rate reduction achieved more about 24.4% for CIF sequence and 18.08% for
20 25 30 35 400
50
100
150
200
250
300
350
400Bitrate vs QP Akiyo Qcif
Bit
rate
(k
bp
s)
Quantization Parameter (QP)
Bitrate Akiyo Qcif Proposed Method
Bitrate Akiyo Qcif JM 18.6
23 24 25 26 27 28 29 30 31
800
1000
1200
1400
1600
1800
2000
2200
2400
Bitrate vs QP tempete cif
Bitra
te (
kbps)
Quantization Parameter (QP)
Bitrate tempete cif Proposed Method
Bitrate tempete cif JM 18.6
0 50 100 150 200 250 300 350 400 450 500 55020
25
30
35
40
45
50Rate Distortion Foreman qcif
Peak S
ignal to
Nois
e R
atio (
dB
)
Bit rate(kbps)
RD curve Foreman qcif Proposed Method
RD curve Foreman qcif JM18.6
-
International Journal of Modern Trends in Engineering and Research (IJMTER) Volume 02, Issue 06, [July– 2015]ISSN (Online):2349–9745 ; ISSN (Print):2393-8161
@IJMTER-2015, All rights Reserved 466
QCIF sequence compared with previous algorithm.PSNR is consistent in the proposed method but
PSNR is not consistent in the prevous algorithm.
From the ratedistortion curve is observe that PSNR is consistent. From the bit rate curve it observe that
bit rate reduces faster in proposed algorithm.
Fig.11.Reconstucted Foreman Intra frame cif
Fig.12. Reconstrcucted tempete frame cif
X. CONCLUSIONS The proposed work is Intra frame coding in AVC using Gaussian pulse for 4x4 intra DC prediction
mode. It is observe that using Gaussian pulse can achieved consistent PSNR and averagely 24.4%
reduced in bit rate for CIF sequence and 18.08% for QCIF sequence.The simulated results are presented
in table (I-VIII) and rate distortion curve, and bit rate curve are represented. The results shows that
consistent reconstructed picture quality (PSNR), reduced in bit rate achieved.
REFERENCES [1] Iain E.Richardson, “The H.264 and MPEG-4 Video Compression: Video coding for Next-generation Multimedia”, Johan
Wiley& Sons, first edition 2003.
-
International Journal of Modern Trends in Engineering and Research (IJMTER) Volume 02, Issue 06, [July– 2015]ISSN (Online):2349–9745 ; ISSN (Print):2393-8161
@IJMTER-2015, All rights Reserved 467
[2] Iain E. Richarson, The H.264 Advanced Video Compression Standard, Johan Wiley& Sons, Second edition 2010.
[3] Chaminda Sampath Kannangara, Complexity Management of H.264/AVC Video Compression, the Robert Gordon
University 2006.
[4] Thomas Wiegand, Gory. Sullivan, Seniour Member, IEEE, Gisle Bjontegaard and Ajay Luthra, “Overview of the
H.264/AVC Video Coding Standard”, IEEE Transactions on circuits and systems for video Technology, Vol. No 7 2003.
[5] Rein van den Boomgaard and Rik van der Weij, “Gaussian Convolutions Numerical Approximations Based on
Interpolation”, Intelligent Sensory Information Systems, University of Amsterdam, and The Netherlands. [6] Pascal Gwosdek, Sven Grewenig1, Andr´es Bruhn, and Joachim Weickert, “Theoretical Foundations of Gaussian
Convolution by Extended Box Filtering”.
[7] Huang Hui, Cao Tie-Yong,Zhang Xiong-Wei, “The enhanced intra prediction algorithm for H.264”IEEE 978-0-7695-
3119-9/2008 Congress on Image and Signal Processing.
[8] Jijun Shi, Yunhui Shi, Baocai Yi, “ Fast intra prediction mode decision for AVC based on edge direction detection” 2008
The Institution of Engineering and technology.
[9] K.Bharanitharam, An-Chao Tsai, “Efficient Block Size Decision Algorithmm for intra mode decision in AVC encoder”
2009 IEEE International Symposium on Multimedia
[10] http://www.vcodex.com