Concealment-Aware Motion Estimation and Mode Selection for Error Resilient Video Coding

This paper proposes a new concealment-aware motion estimation and mode selection scheme for error resilient video coding, where a new distortion metric is adopted to better accounts for the impact of "worst-case" distortion on the overall perceptual video quality than the existing statistically defined end-to-end distortion metric does. Effective "motion copy" error concealment is assumed at the decoder. The proposed scheme requires no end-to-end distortion estimation, and thus, achieves low complexity. Simulation result with an H.264 codec shows that our proposed scheme improves subjective decoder reconstructed video quality, especially for sequences of medium motion.