An improved inter frame error concealment in H.264/AVC

Jun Wang*, Lei Wang, Shen Li, Takeshi Ikenaga, Satoshi Goto

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

Transmission of compressed video over error prone channels may result in packet losses or errors, which can significantly degrade the image quality. Aimed this problem, non-normative error concealment (EC) is recommended in H.264/AVC to recover the lost image. By analyzing of non-normative inter EC, this paper focused on improvement of it. Firstly, since the "guessed"/ recovered MV (motion vector) in existing EC is not so close to the real MV, a refined MV recovery is proposed to get a more accurate "guessed" MV. Secondly an improved MV selection criterion (side matching criterion) is proposed, which is trying to "guess"/approach a lost MB (macroblock)'s real/original side match distortion, instead of "guess"/approach value 0 in traditional inter EC. In addition, for MV refinement scheme, inspired by MV searching in motion estimation, we adopt a diamond searching for MV recovery. Both objective and subjective image quality evaluation in experiments show that our proposal achieves a better image recovery compared with non-normative inter EC.

Original languageEnglish
Title of host publicationASICON 2007 - 2007 7th International Conference on ASIC Proceeding
Pages794-797
Number of pages4
DOIs
Publication statusPublished - 2007 Dec 1
Event2007 7th International Conference on ASIC, ASICON 2007 - Guilin, China
Duration: 2007 Oct 262007 Oct 29

Publication series

NameASICON 2007 - 2007 7th International Conference on ASIC Proceeding

Conference

Conference2007 7th International Conference on ASIC, ASICON 2007
Country/TerritoryChina
CityGuilin
Period07/10/2607/10/29

ASJC Scopus subject areas

  • Hardware and Architecture
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'An improved inter frame error concealment in H.264/AVC'. Together they form a unique fingerprint.

Cite this