An Efficient Encoder Rate Control Solution for Transform Domain Wyner-Ziv Video Coding
Brites , C.
IEEE Transactions on Circuits and Systems for Video Technology Vol. 21, Nº 9, pp. 1278 - 1292, September, 2011.
ISSN (print): 1051-8215
Journal Impact Factor: 2,951 (in 2008)
Digital Object Identifier: 10.1109/TCSVT.2011.2147210
Download Full text PDF ( 618 KBs)
Most Wyner-Ziv (WZ) video coding solutions in the literature use a feedback channel (FC) based decoder rate control (DRC) strategy to adjust the bitrate to correct the side information (SI) errors. More recently, some encoder rate control (ERC) strategies have been proposed to address application scenarios where a FC is not available. The ERC based WZ video coding RD performance depends not only on the (encoder) parity rate estimator (PRE) accuracy but also on the decoder ‘intelligence’ in dealing with the residual errors due to parity rate underestimation. In this context, the main objective of this paper is to propose a more efficient and powerful ERC solution for transform domain WZ (TDWZ) video coding by simultaneously tackling the two issues aforementioned with the following technical novelty: 1) integration in an ERC context of Gray mapping for the quantized DCT coefficients to enhance the correlation between WZ and SI data; 2) more accurate PRE to better estimate the needed parity rate to avoid undesired parity rate underestimations and overestimations; 3) novel soft reconstruction function to reduce the impact of the residual bitplane errors in the decoded WZ frame quality; and 4) weighted overlapped block motion compensation technique to refine the SI used in an iterative WZ decoding framework with the correlation noise model parameters dynamically updated. Experimental results show a considerable RD performance improvement with a reduction of up to about 2 dB of the gap between the ERC and DRC based approaches in TDWZ video coding solutions, thus making this ERC based WZ codec the most efficient available and competitive regarding DRC based WZ video coding solutions.