H.264 STANDARD BASED SIDE INFORMATION

GENERATION IN WYNER-ZIV CODING

Subrahmanya M V(Under the guidance of Dr. Rao and Dr.Jin-

soo Kim)

Contents

Traditional Video coding

Wyner-Ziv video coding

Applications

Wyner-Ziv encoder◦ Key frame encoding◦ WZ frame encoding

Wyner-Ziv decoder◦ Key frame decoding◦ WZ frame decoding

Contents( contd.)

Side Information Generation◦ Inter-prediction◦ Intra-prediction

Motion estimation (ME)◦ Full-pixel◦ Half-pixel◦ Quarter-pixel

Motion compensation (MC)◦ Luma◦ Chroma

Bi-directional interpolationIntra-prediction

Contents( contd.)

ImplementationTestingResults

◦SI prediction schemes◦ME block sizes◦ME search ranges◦Key frame distances◦Bit rates

ConclusionsReferences

Traditional Video coding

Source coding Encoder removes temporal and spatial redundancies

Intra frame encoding Inter frame encoding: Computationally intensive

Complex Encoder + Simple decoder

[Dr. Jin-soo Kim]

Wyner-Ziv video coding

Source coding + Channel coding Decoding with side-information(SI)

Slepian-Wolf coding: Lossless Wyner-Ziv coding: Lossy

Simple encoder + Complex decoder

[Dr. Jin-soo Kim]

Applications

Light encoder and light decoder [20] Wireless low power video surveillance Sensor network Multi-view image acquisition Networked camcoders

[Dr. Jin-soo Kim]

Wyner-Ziv Encoder

Input YUV 420 Output: Bitstream Encoding Key frame + WZ frame

Key frame encoding

H.264 Intra frame coding Intra prediction Integer Transform Quantization Entropy Coding

[M. Lee, A. Moore]

Intra prediction

Predicted from top, left, top left and top right

neighbors 4x4 intra prediction Nine directional modes

[J. S. Park and H. J. Song]

[T. Wiegand and G.J Sullivan]

Key frame encoding

Integer Transform

Quantization

◦ Element by element Multiplication

Entropy Coding using Exp-Golomb codes

WZ frame encoding

Residual signal Generation

-26 -19 -12 -5 2 9 16 23 30 37 44 51 58 65 720

0.010.020.030.040.050.060.070.080.090.1

ForemanStefancoastguard

[Dr. Jin-soo Kim]Residual Signal

WZ frame encoding(contd.)

Residual bit plane extraction

WZ frame encoding(contd.)

Low Density Parity Check(LDPC) encoding Obtain parity bits for the message bits Message bits(90%) + Parity bits(10%) Only parity bits are sent to achieve compression

Wyner-Ziv Decoder

Decoding Key frame + WZ frame

[E. Peixoto, R. L. de Queiroz, and D. Mukherjee]

Key Frame

WZ Frame

Wyner-Ziv Decoder( Contd. )

Key frame decoding Intra prediction Entropy decoding Inverse Quantization Inverse Integer transform

Key frame decoding

Inverse Integer Transform

Inverse Quantization

◦ Element by Element Multiplication

Entropy decoding Exp-Golomb codes are decoded using prefix tables

Wyner-Ziv Decoder( Contd. )

WZ frame decoding SI generation: Prediction to obtain the message bits LDPC decoding: Correct prediction errors using parity bits

Message contains a pair of values [qij(0), qij(1)] or [rij(0), rij(1)] – Signify the amount of belief that yi is a ”0” or a ”1”.

Initial belief is based on message bits predicted and parity bits sent Belief is iteratively updated and new message bits are obtained Iteration proceeds till fixed iteration count or message bits remain

unchanged in an iteration

Side Information Generation

Motion estimation Estimate motion of a block between two key frames

to obtain motion vectors( MVx, MVy ) for each block Interpolate motion vectors to obtain motion

between key frame and WZ frame

Motion Compensation Interpolate key frame based on motion vectors to

obtain MC prediction for WZ frame

Intra prediction Prediction for blocks which are absent in key frames

Motion Estimation

Forward ME between key frames – MVF

Backward ME between key frames - MVB

Derive MVs for WZ frame using Key frame MVS

[E. Peixoto, R. L. de Queiroz, and D. Mukherjee 2008]

Higher key frame distance

For larger key frame distance hierarchical structure can be used

[A. Aaron, E. Setton and B. Girod 2003]

SAD based ME

Obtain absolute difference between each pixel in the block to be searched and the reference block

Best match is a block with least sum of absolute differences

(W+2SR) X (H+2SR)

Sub-pixel ME

Motion vector is estimated up to quarter-pixel units

Interpolation of reference is done to obtain half-pixel and quarter-pixel units

Half-pixel positions are obtained using five tap filter

Quarter pixels are obtained using bilinear interpolation

Half-pixel motion estimation

H33 = [F13 + -5 * F23 + 20 * F33 + 20 * F43 + -5 * F53 + F63 + 15] >> 5

G33 = [F31 + -5 * F32 + 20 * F33 + 20 * F34 + -5 * F35 + F36 + 15] >> 5

D33 = [H31 + -5 * H32 + 20 * H33 + 20 * H34 + -5 * H35 + H36 + 15] >> 5

quarter-pixel motion estimation

q1 = ( F33 + G33 + 1 ) >> 1

q2 = ( G33 + F34 + 1 ) >> 1

q3 = ( F33 + H33 + 1 ) >> 1

q4 = ( H33 + G33 + 1 ) >> 1

q5 = (G33 + D33 + 1 ) >> 1

q6 = ( G33 + H34 + 1 ) >> 1

q7 = ( H33 + D33 + 1 ) >> 1

q8 = ( D33 + H34 + 1 ) >> 1

q9 = ( H33 + F43 + 1 ) >> 1

q10 = ( H33 + G43 + 1 ) >> 1

q11 = ( D33 + G43 + 1 ) >> 1

q12 = ( G43 + H34 + 1 ) >> 1

Motion Compensation (Luma)

Obtain a block from key frame with offset specified by motion vector

Motion Compensation (Chroma)

Chroma MV = Luma MV / 2 Obtain A, B, C, D based on integer part of MV

Floor(MV/8)

Predict based on fractional part of MV (MV & 7)

Bi-directional interpolation

SI frame = (Forward prediction Frame + Backward Prediction frame + 1)/2

Intra prediction

4x4 prediction Prediction mode of co-located block from

previous key frame

Implementation

EncoderKey frame encoding is done using JM encoderWZ frame decoding is implemented in ‘C’

Residual generation Quantization Bit plane extraction LDPC encoder

DecoderKey frame is decoded using JM decoderWZ frame decoding is implemented in ‘C’

SI generation LDPC decoder De-quantization Reconstruction

LDPC encoder and decoder are implemented by Dr. Jin-soo Kim [4]

Testing

Comparison with traditional codingH.264 is used for comparison Generate H.264 reconstructed sequence

Encode using H.264 encoder Decode using H.264 decoder to obtain reconstructed

YUV sequenceGenerate Wyner-Ziv reconstructed sequence

Encode using Wyner-Ziv encoder Decode using Wyner-Ziv decoder to obtain

reconstructed YUV sequenceCompare two sequences using PSNR and SSIM

Significance of different parameters in Wyner-Ziv coding SI Prediction methodKey frame distanceME search rangeME block size

SI prediction schemes(Sample reconstructed images)

bus( CIF) [Average, Full pixel ME, Half-pixel ME, Quarter-pixel ME, Bi-prediction, Intra-prediction combined]

SI prediction schemes( Plots)

ME block sizes(Sample reconstructed images)

coastgaurd (CIF) [16x16, 8x8, 4x4]

ME block sizes( Plots)

ME search ranges (Sample reconstructed images)

stefan (CIF) [16, 8, 4]

ME search ranges( Plots)

Key frame distances (Sample reconstructed images)

garden (CIF) [1, 2, 3, 4, 5]

Key frame distances( Plots)

Comparison with H.264 (Sample reconstructed images)

Comparison with H.264( Plots)

Conclusions

H.264 based SI generation based on quarter-pixel ME performs better compared to previously done H.263which is based on half-pixel ME[1] .

ME block size 16x16 is ideal

Higher ME search range produces better results

At higher key frame distance quality drops sharply

Traditional video coding has better R-D performance

Future work

Wyner-Ziv to H.264 transcoder Light encoder + light decoderInter and intra prediction done in Wyner-

Ziv decoder can be reused for the H.264 encoder

Accepted conference paper

Subrahmanya Venkatrav and Dr. K. R. Rao,

“Side information generation in Wyner-Ziv

decoder”,

The 7th International Joint Conference on Computer Science and Software

Engineering

May 13-14, 2010 Bangkok, Thailand

References

1. E. Peixoto, R. L. de Queiroz, and D. Mukherjee, “Mobile video communications using a Wyner-Ziv transcoder,” Proc. SPIE 6822, VCIP, 68220R Jan. 2008.

2. A. Aaron, E. Setton and B. Girod, "Towards practical Wyner-Ziv coding of video," Proceedings. IEEE International Conference on Image Processing, 2003. ICIP 2003., vol.3, pp. III-869-872, 14-17 Sept. 2003.

3. K. R. Rao and J. J. Hwang, Techniques and standards for image, video, and audio coding, Prentice Hall PTR, 1996.

4. Jin-Soo Kim, "Brief overview of Wyner-Ziv CODEC" (Private Communication)

5. A. Aaron, D. Varodayan, and B. Girod, “Wyner-Ziv residual coding of video,” Proc. International Picture Coding Symposium, Beijing, P. R. China , April 2006.

6. T. Wiegand and G.J Sullivan, “The H.264/AVC video coding standard”, IEEE SP Magazine, vol. 24, pp. 148-153, March 2007.

7. G. J. Sullivan, P. Topiwala, and A. Luthra, "The H.264/AVC advanced video coding standard: Overview and introduction to the fidelity range extensions", SPIE Conf. on applications of digital image processing XXVII, vol. 5558, pp. 53-74, Aug. 2004.

8. S.K. Kwon, A. Tamhankar, and K.R. Rao “Overview of H.264/MPEG-4 Part 10” J. VCIR, Vol. 17, pp. 186-216, April 2006, Special Issue on “Emerging H.264/AVC Video Coding Standard,”.

References(Contd)

9. A. Wyner and J. Ziv, "The rate-distortion function for source coding with side information at the decoder,"  IEEE Trans., Information Theory, vol.22, pp. 1-10, Jan 1976.

10. D. Slepian and J. Wolf, “Noiseless coding of correlated information sources,” IEEE Trans. on Information Theory Vol. 19, pp. 471–480, July 1973.

11. D. Varodayan, A. Aaron and B. Girod, "Rate-adaptive distributed source coding using low-density parity-check codes,"  Conference Record of the Thirty-Ninth Asilomar Conference on Signals, Systems and Computers, pp. 1203-1207, Oct. 28 – Nov. 1, 2005.

12. Z. Li and E.J. Delp, "Wyner-Ziv video side estimator: conventional motion search methods revisited," IEEE International Conference on Image Processing, 2005. ICIP 2005, vol.1, pp. I-825-828, 11-14 Sept. 2005.

13. L. Liu and E. J. Delp, "Wyner-Ziv video coding using LDPC codes," Proceedings of the 7th Nordic Signal Processing Symposium, 2006. NORSIG 2006.

14.  D. Kubasov, K. Lajnef and C. Guillemot, "A hybrid encoder/decoder rate control for Wyner-Ziv video coding with a feedback channel," IEEE 9th Workshop on Multimedia Signal Processing, 2007. MMSP 2007., pp.251-254, 1-3 Oct. 2007.

15. C. Brites and F. Pereira, "Encoder rate control for transform domain Wyner-Ziv video coding," IEEE International Conference on Image Processing, 2007. ICIP 2007., vol.2, pp.II -5-II -8, 16-19 Sept. 2007

16. A. Roca, et al, "Rate control algorithm for pixel-domain Wyner-Ziv video coding ," Proc. SPIE, vol. 6822, 68221T (2008).

References(Contd)

17. D. Mukherjee, “Optimal parameter choice for Wyner-Ziv coding of Laplacian sources with decoder side information,” HP Labs Technical Report HPL-2007-34, 2007.

18. Z. Wang, et al, "Image quality assessment: From error visibility to structural similarity," IEEE Trans., Image Processing, vol.13, pp. 600-612, April 2004

19. I. Richardson, “The H.264 advanced video compression standard,” Hoboken, NJ: Wiley, 2010.

20. D. Rebollo-Monedero, S. Rane, A. Aaron and B. Girod, "High-rate quantization and transform coding with side information at the decoder," EURASIP Signal Processing Journal, Special Issue on Distributed Source Coding.

21. S.-K Kwon, A. Tamhankar and K.R. Rao, ‘Overview of H.264 / MPEG-4 Part 10” ISME, Hong Kong, Oct. 2004.

22. JVT documents ftp://standards.polycom.com

http://ftp3.itu.ch/av-arch/jvt-site/23. M. Lee and A. Moore, “H.264 Encoder Design“, Group 3, May 17,

200624. J. S. Park and H. J. Song, “Selective Intra Prediction Mode Decision

for H.264 /AVC Encoders”, World Academy of Science, Engineering and Technology 13 2006.

25. B. M.J. Leiner, “LDPC Codes – a brief Tutorial”, April 8, 2005

References(Contd)

26. G. Cote, B. Erol, M. Gallant, and F. Kossentini, “H.263+: Video coding at low bit rates,” IEEE Trans. Circuits and Systems for Video Technology ,Vol.8, pp. 849–866, Nov. 1998

27. (2008, August) H.264/avc JM reference software. Joint Video Team (JVT) of ISO/IEC MPEG & ITU-T VCEG. [Online]. Available: http://iphome.hhi.de/suehring/tml/