DocumentCode :
3094854
Title :
Overlapping Interval Differences-Based Fast Intra Mode Decision for H.264/AVC
Author :
Lei, Yan-Qiang ; Wang, Yuan-Gen ; Liang, Fan
Author_Institution :
Dept. Electron. & Commun. Eng., Sun Yat-sen Univ., Guangzhou, China
fYear :
2010
fDate :
15-17 Oct. 2010
Firstpage :
659
Lastpage :
663
Abstract :
An efficient and fast mode decision algorithm for H.264/AVC Intra prediction is proposed in this paper. Overlapping interval differences (OID) are calculated in the domain composed of samples of current block and reconstructed samples of neighboring blocks. The OID can represent the direction error of each prediction mode. Smaller direction error means that the corresponding prediction mode will be considered as the final selection with high probability by rate-distortion optimization (RDO). So this method adopts a subset of the candidate modes instead of all to implement the efficient and fast encoding. Experimental results show that the proposed scheme can reduce about 30% computational complexity with negligible degradation of video quality and compression efficiency. Compared with other approaches in the literature, the proposed algorithm also shows better performance.
Keywords :
computational complexity; data compression; optimisation; video coding; H.264/AVC intraprediction; computational complexity; encoding; fast intramode decision algorithm; overlapping interval differences; rate distortion optimization; video quality; Algorithm design and analysis; Automatic voltage control; Encoding; PSNR; Pixel; Prediction algorithms; Signal processing algorithms; H.264/AVC; Intra prediction; direction error; fast mode decision; overlapping interval difference;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Intelligent Information Hiding and Multimedia Signal Processing (IIH-MSP), 2010 Sixth International Conference on
Conference_Location :
Darmstadt
Print_ISBN :
978-1-4244-8378-5
Electronic_ISBN :
978-0-7695-4222-5
Type :
conf
DOI :
10.1109/IIHMSP.2010.167
Filename :
5636207
Link To Document :
بازگشت