A fast feature-assisted adaptive early termination approach for multiple reference frames motion estimation in H.264

Abstract The multiple reference frames motion estimation approach used in H.264 is computationally intensive. This paper presents a fast or computationally efficient feature-assisted adaptive early termination approach in order to reduce the computational complexity while maintaining more or less the same video quality. The introduced feature-assisted approach consists of three parts: (1) reduction of the number of available reference frames using predicted motion activity, extracted texture information, and skip mode from neighboring macroblocks, (2) the most probable reference frame prediction based on neighboring macroblocks, and (3) an adaptive early termination threshold derived from a theoretical analysis of all zero block detection. Extensive experimental results are performed to demonstrate the computational gain of the introduced approach over the standard approach for the multiple reference frames motion estimation. Keywords Multiple reference frames motion estimation

• Type: Journal
• Year: 2008
• Where: JRTIP
• Authors: Jianfeng Ren, Nasser D. Kehtarnavaz, Madhukar Budagavi
• Keywords: Frames Motion Estimation, JRTIP 2008, Reference Frames, Reference Frames Motion