Rate distortion-based motion estimation search ordering for rate-constrained successive elimination algorithms

Abstract

In this paper, we propose a new class of search ordering algorithms to reduce the computational cost of motion estimation in video coding. We show that conventional search orderings, such as spiral search, can weaken the filtering criterion of rate-constrained successive elimination algorithms. Based on this new insight, we derive a new search ordering that takes into account the impact of the rate constraint. Our simulation results demonstrate that, on average, the amount of SAD operations required to encode the tested sequences, is reduced by 2.86%, when compared to the H.264 JM reference software's implementation of spiral search. For sequences with unpredictable motion, this reduction is greater than 5% and can exceed 10% when smaller block partitions are evaluated.

An adaptive search ordering for rate-constrained successive elimination algorithms

“…For example, a spiral search ordering applied to a bad predicted motion vector would cause multiple unnecessary cost function evaluations. Even with a good motion vector prediction, the spiral ordering does not follow the monotonically increasing rate rule defined in [12]. This will lead to unnecessary cost function evaluations as the SEA threshold in equation 2 will vary according to the difference in rate between the best block matching candidate and the current candidate.…”

“…This will lead to unnecessary cost function evaluations as the SEA threshold in equation 2 will vary according to the difference in rate between the best block matching candidate and the current candidate. While exponential Golomb codes are no longer used to encode motion vectors, the method proposed by Trudeau et al [12] for H.264 can be extended to HEVC. In HEVC, the bit length of the exponential Golomb encoding of the motion vector is recommended as a fast estimate of the motion vector cost for the rate-constrained motion estimation algorithm (as seen in the HEVC reference software [10]).…”

“…The new class of search orderings, proposed in [12], also known as rate-constrained search orderings, are based on the monotonically increasing rate rule…”

“…However, as described in [12], this leads to unnecessary cost function evaluations. To resolve this problem, we propose an approach to adapt the search ordering of block matching candidates.…”

“…Spiral search ordering is considered state-ofthe-art, and is used in many implementations of SEA-based algorithms [8,9,11]. However, Trudeau et al [12] showed that a spiral search ordering, in a rate-constrained context, will perform unnecessary cost function evaluations. To avoid this, the motion vector cost of the search ordering must increase monotonically, otherwise the SEA filtering criterion is weakened.…”