Skip to Main Content (Press Enter)

Logo CNR
  • ×
  • Home
  • People
  • Outputs
  • Organizations
  • Expertise & Skills

UNI-FIND
Logo CNR

|

UNI-FIND

cnr.it
  • ×
  • Home
  • People
  • Outputs
  • Organizations
  • Expertise & Skills
  1. Outputs

Performance evaluation of FMO to improve the error resilience of H.264/AVC

Academic Article
Publication Date:
2006
abstract:
The transmission of compressed video streams over error-prone networks requires an encoder equipped with error resilience tools. The novel H.264/AVC standard supports several of such tools; in particular, it allows to group the macroblocks of an image in a very flexible way by means of the Flexible Macroblock Ordering (FMO) mapping functions. This feature is very useful in terms of robustness towards errors and error concealment, although it introduces some source coding overhead. In this paper we report a thorough performance assessment of FMO for H.264/AVC video transmission over lossy packet networks. We consider sequences in QCIF and CIF resolution, different slicing options, and different bit-rates. We first investigate the overhead of FMO in case of no transmission errors, and then evaluate its performance for wireless transmission, employing a random packet loss model, as well as UMTS error patterns. An advantage of FMO is that, using the checkerboard-like pattern, it is possible to avoid the loss of several consecutive macroblocks, potentially improving the error concealment performance. We introduce a simple error concealment technique that, unlike the JVT concealment, exploits the FMO structure. Simulation results show that FMO provides a significant performance improvement on error-prone channels, and ad-hoc error concealment can further enhance the end-to-end image quality.
Iris type:
01.01 Articolo in rivista
Keywords:
Error Concealment; Error Resilience; Flexible Macroblock Ordering; H.264/AVC
List of contributors:
Baccichet, Pierpaolo; Chimienti, Antonio
Handle:
https://iris.cnr.it/handle/20.500.14243/12822
Published in:
ST JOURNAL
Journal
  • Use of cookies

Powered by VIVO | Designed by Cineca | 26.5.0.0 | Sorgente dati: PREPROD (Ribaltamento disabilitato)