Efficient index assignment by improved bit probability estimation for parallel processing of distributed video coding

Distributed video coding (DVC) is attracting attention as a paradigm for video compression. One reason is that it transfers the high complexity computation to the decoder. However, the DVC decoder that uses channel coding based on a belief propagation algorithm has complexity exceeding that of the H.264/AVC encoder. This paper proposes a parallelized DVC scheme that treats each bitplane independently. Unfortunately, simple parallelization schemes suffer low compression efficiency because they can't use additional side information for the decoding of subsequent bitplanes. Therefore, we propose an effective estimation method that can calculate the bit probability as accurately as possible by index assignment. Simulations show that the proposed system can reduce the decode time by up to about 30-35 [%] with only slight parallelization loss.