Decoder-side block motion estimation for H.264 / MPEG-4 AVC based video coding

Abstract

In video coding standards like H.264 / MPEG-4 AVC, the encoder performs motion estimation in order to utilise temporal dependencies within a sequence. In addition to the rate of the residue, the encoder has to allocate bits for motion vectors required to compensate the motion at the decoder. This bit rate increases for smaller block sizes, since more motion vectors need to be transmitted. Therefore, motion compensation using dense motion vector field is not feasible for such an architecture.This paper proposes to estimate motion for coding of B frames at the decoder. Using this decoder-side motion estimation, the transmission of the motion vectors is not necessary and the bit rate is reduced. Furthermore, prediction quality is higher in many cases resulting in a coding gain of up to 1.7 dB at low bit rates and 0.2 dB at higher bit rates.

Decoder-Side Motion Estimation Assuming Temporally or Spatially Constant Motion

“…This section explains the architecture used to allow motion estimation at the decoder, which is based on [3]. Additionally, it is shown how nonlinear motion can be compensated, although linear motion is assumed in the decoder-side motion estimation algorithm.…”

“…In addition to this approach with the modified reference list, a pure DSME mode was proposed in [3]. For low bit rates, the interpolated DSME frame F 0 may have sufficient quality and, thus, no additional residual has to be transmitted.…”

“…Thus, the current block is able to follow the motion of every neighbouring block, while still using small search ranges. This significantly reduces the amount of MAD computations compared to the search algorithm used in [3].…”

“…Another approach, discussed in more detail in this paper, is decoder-side motion estimation (DSME, [3]). Hereby, a prediction of the current frame is generated by performing bidirectional interpolation using previously coded frames.…”

Motion Estimation at the Decoder

“…The system was later extended to support DMVD also within B macroblock modes (Kamp & Wien, 2010). Decoder-side motion estimation (DSME), proposed by Klomp et al (2009), is another approach to reduce the rate for the motion vectors. Temporal interpolation is used to generate a prediction of the current frame at the decoder.…”