Error resilient motion estimation in multiple reference frame video coding

  • Authors:
  • Hui-Seon Gang;Jae-Young Pyun;Suk-Seung Hwang

  • Affiliations:
  • Chosun University, Gwangju, Korea;Chosun University, Gwangju, Korea;Chosun University, Gwangju, Korea

  • Venue:
  • Proceedings of the 8th International Conference on Ubiquitous Information Management and Communication
  • Year:
  • 2014

Quantified Score

Hi-index 0.00

Visualization

Abstract

H.264/AVC is a video coding standard widely used in many multimedia services because of its higher coding efficiency and various application possibilities. In order to increase the compression ratio, H.264/AVC standard uses multiple reference frames for the motion estimation and decides the optimal motion vectors and encoding mode by Rate-Distortion Optimization (RDO). However, using multiple reference frames causes to degradation of video quality due to the ineffectiveness of error resilience and error propagation when some errors occurs in video data during transmission. In this paper, we show analysis of error resilience algorithms with respect to the multiple reference frames and introduce a new error resilience algorithm providing more robustness to errors. The proposed error resilience algorithm makes encoder select an optimal reference block considering both intra-refresh and RDO based cost between original and reference blocks. In addition, experimental results demonstrate that the proposed error resilience is effective even in the presence of multiple reference frames. The performance gain is achieved by choosing blocks satisfying with both coding efficiency and error robustness during motion estimation at the cost of bit rate overhead.