1. C.K. Kim, D.Y. Suh, J. Park, B. Jeon ha 強壯 !

2. DVC bitstream reorganiser

3.  Mobile-to-mobile video communications require both an encoder and a decoder to have low complexity.  High complexity at the DVC decoder side makes it difficult for resource-constrained mobile devices to perform decoding. ha 強壯 !

4.  DVC bitstream reorganiser estimates block- wise motion vectors and sends the motion information and the parity data required for decoding to the decoder ha 強壯 !

5.  DVC encoder (block A)  DVC bitstream reorganiser (block B)  DVC decoder (block C) ha 強壯 !

7.  In the proposed scheme, Wyner-Ziv decoder compensates wrong blocks by side information using side matching and bi- directional searching. The noise reduction in side information allows the proposed algorithm to achieve coding improvements not only in bit-rate but also in PSNR. Results of our experiments show improvement of PSNR up to 0.4dB. ha 強壯 !

8.  transform domain Wyner-Ziv (TDWZ)  LDPCA  side matching method(SI)  bitstream reorganiser ha 強壯 !

9.  Conventional DVC decoder  Parity amount estimator  Side information renovator ha 強壯 !

10.  Conventional DVC encoder ha 強壯 !

11.  Side information renovator(ME)  Parity amount estimator ha 強壯 !

12.  Generates side information using decoded KEY and WZ pictures.  The decoded WZ pictures exploited here to generate side information enhance the quality of that information. ha 強壯 !

13.  A precise estimation of the parity data required for decoding is possible after Slepian-Wolf decoding has been performed. ha 強壯 !

14.  Eventually, the estimated amount of parity bits and motion information obtained in the generation of side information are delivered to the DVC decoder at the target device. ha 強壯 !

15.  The WZ picture decoder (block C 2 ) decoding is faster. ha 強壯 !

16.  This is because the decoded KEY pictures and motion information passed from the DVC bitstream reorganiser (block B) enable it to generate the side information without performing motion estimation. ha 強壯 !

17.  Increasing the decoding speed  The side information generated at the DVC decoder (block C) is as good as the information generated in the side information renovator (block B 3 ) in terms of Quality ha 強壯 !

18.  The amount of parity bits required for Slepian-Wolf decoding is fed into the WZ picture decoder (block C 2 ) at this time.  The removal of iterative channel decoding contributes to reducing computational cost. ha 強壯 !

20.  DVC bitstream reorganiser which reduces decoder complexity so that decoder speed becomes 10 to 25times faster.  The DVC bitstream reorganiser sends the motion information and parity bits required for DVC decoding to the decoder at the target device. ha 強壯 !

21.  Mobile to Mobile Video Communication using DVC Framework with Channel Information Transmission ha 強壯 !

23.  However it can have additional quality degradation coming from re-quantization ha 強壯 !

24.  However, as motion vector information and parity length information increases, its RD performance can be degraded. ha 強壯 !

26.  CNM1 estimates the Laplace distribution parameter for each block ◦ It only uses already decoded key pictures  CNM2 also uses decoded Wyner-Ziv picture information ◦ globally estimate it. Therefore, performance by CNM2 is limited  we selectively use CNM1 and CNM2 by considering the bit rate. ha 強壯 !

27.  It generates parity bit-stream by using  DCT transform, quantization [6,7], and rate-adaptive LDPC  accumulate (LDPCA) coding [8]. ha 強壯 !

28.  Decoder produces the side  information by using the method in [9] which exploits template  matching processing. ha 強壯 !

29.  8 bit cyclic redundancy check (CRC) data  for each bit-planes help to detect the LDPCA decoding error  [10]. DCT coefficients are reconstructed by using [11]. ha 強壯 !

37.  Correlation Noise Modeling for Efficient Pixel and Transform Domain Wyner–Ziv Video Coding ha 強壯 !

38.  9/14->10/14 ◦ mobile to PC  10/14->11/14 ◦ Mobile to mobile ha 強壯 !

39.  Thank you ha 強壯 !