Title :
Bi-direction Motion Vector retrieval based error concealment scheme for H.264/AVC
Author :
Tang, Yichun ; Wang, Jun ; Goto, Satoshi
Author_Institution :
Grad. Sch. of Inf. Production, & Syst. LSI, Waseda Univ., Kitakyushu
Abstract :
As the newest video coding standard, H.264/AVC adopts the high-efficiently predictive coding and variable length entropy coding to achieve high compression efficiency. On the other side, transmission errors become the major problem faced by video broadcasting service providers. Error concealment (EC) here is adopted to handle slices with huge conjunctive corrupted areas inside. Considering error propagation from corrupted slice to succeeding ones is the key factor affecting the video quality, this paper proposes a novel temporal EC scheme including the bi-direction motion vector (MV) retrieval method and an adaptive EC ordering basing on it. Background and motional steady shift part of slice will be given top and second priority, respectively. Combined with our proposed improved boundary matching algorithm (IBMA) which provides more accurate distortion function, experiments results show that our proposal achieves better performance under different error rate channel, compared with EC algorithm adopted in H.264 reference software.
Keywords :
code standards; entropy codes; error statistics; image motion analysis; image retrieval; video coding; H.264/AVC; IBMA; bidirection motion vector retrieval; compression efficiency; error concealment; error concealment scheme; error rate channel; high-efficiently predictive coding; improved boundary matching algorithm; transmission errors; variable length entropy coding; video broadcasting service providers; video coding standard; video quality; Automatic voltage control; Bidirectional control; Broadcasting; Entropy coding; Multimedia communication; Predictive coding; Proposals; Software algorithms; Video coding; Video compression;
Conference_Titel :
Signal Processing & Its Applications, 2009. CSPA 2009. 5th International Colloquium on
Conference_Location :
Kuala Lumpur
Print_ISBN :
978-1-4244-4151-8
Electronic_ISBN :
978-1-4244-4152-5
DOI :
10.1109/CSPA.2009.5069214