Heegard-Berger video coding using LMMSE estimator

Xiaopeng Fan,Oscar Au,Yan Chen,Jiantao Zhou
DOI: https://doi.org/10.1007/11922162_15
2006-01-01
Abstract:In this paper a novel distributed video coding scheme was proposed based on Heegard-Berger coding theorem, rather than Wyner-Ziv theorem. The main advantage of HB coding is that the decoder can still decode and output a coarse reconstruction, even if side information degrade or absent. And if side information present or upgrade at decoder, a better reconstruction can be achieved. This robust feature can solve the problem lies in Wyner-Ziv video coding that the encoder can hardly decide the bit rate because rate-distortion was affected by the side information known only at the decoder. This feature also leaded to our HB video coding scheme with 2 decoding level of which we first reconstruct a coarse reconstruction frame without side information, and do motion search in previous reconstructed frame to find side information, then reconstruct a fine reconstruction frame through HB decoding again, with side information available.
What problem does this paper attempt to address?