Cross-Hexagon-based motion estimation algorithm using motion vector adaptive search technique

Abstract

This paper proposes a novel cross-Hexagon-based motion estimation algorithm using motion vector adaptive search technique. The proposed algorithm makes full use of the motion vector's characteristics which are center-biased and the relevance of time and space. The proposed algorithm is combined with the cross pattern and hexagon search algorithm, according to the type of motion to adaptively choose the search strategy and the original searching point, at the same time setting the threshold for stationary block to directly suspend searching. The experimental results show that the speed of the search algorithm is superior to the existing fast motion estimation algorithm, searching accuracy close to UMHexagons algorithm.