Dynamic Reference Frame Selection for Improved Motion Estimation Time in H.264/AVC

Multiple reference frames is adopted in H.264/AVC to improve coding efficiency. However it also increases the complexity of motion estimation proportional to the number of reference frames. In this paper, we propose an efficient reference frame selection algorithm to decrease the complexity of motion estimation. By selecting an optimal reference frame we reduce the number of reference frames to be investigated. Also, considering additional reference frames is used to avoid local minima. The proposed algorithm is implemented on the H.264/AVC encoder. The experimental results show that the proposed algorithm decreases video encoding time, while maintaining quality.