DocumentCode
432509
Title
Hierarchical mode search with classification of bisectional prediction modes based on the position of motion boundary [video coding]
Author
Kato, Shigeo ; Sugimoto, Kuzuo ; Moschetti, Fulvio ; Boon, Choong Seng
Author_Institution
Multimedia Labs., NTT DoCoMo, Inc, Kanagawa, Japan
Volume
1
fYear
2004
fDate
24-27 Oct. 2004
Firstpage
485
Abstract
Video coding schemes, equipped with various kinds of prediction modes, improve coding efficiency by offering efficient prediction modes that adapt to the motion in a macroblock (MB). Operation time for motion estimation and mode selection, on the other hand, increases with the number of prediction modes. We have proposed a motion compensated prediction method (arbitrary-shape partitioning motion compensation: APMC) equipped with various inter prediction modes which adapts to the motion boundary of objects on an MB basis. This paper confirms the coding efficiency of APMC in comparison to the final specification of H.264/AVC and introduces a method to improve the computational performance of APMC by using a hierarchical mode search method based on motion boundary position within MBs. Simulation results show that APMC improves coding efficiency by about 5% in comparison with the final specification of H.264/AVC, and the proposed hierarchical mode search method decreases the operation time to about 60%, while maintaining the coding efficiency.
Keywords
adaptive codes; motion compensation; motion estimation; video coding; APMC; H.264/AVC; arbitrary-shape partitioning motion compensation; bisectional prediction mode classification; coding efficiency; hierarchical mode search; inter prediction modes; macroblock motion adaptive coding; mode selection; motion boundary position; motion compensated prediction method; motion estimation; video coding prediction modes; Automatic voltage control; Computational modeling; Laboratories; MPEG 4 Standard; Motion compensation; Motion estimation; Prediction methods; Rate-distortion; Search methods; Video coding;
fLanguage
English
Publisher
ieee
Conference_Titel
Image Processing, 2004. ICIP '04. 2004 International Conference on
ISSN
1522-4880
Print_ISBN
0-7803-8554-3
Type
conf
DOI
10.1109/ICIP.2004.1418796
Filename
1418796
Link To Document