Share Email Print

Proceedings Paper

All-zero block detection for H.264 intra prediction
Author(s): Jianning Zhang; Lifeng Sun; Yuzhuo Zhong
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

H.264/MPEG-4 AVC video coding achieves high coding performance by complex inter and intra prediction algorithms. Many fast intra prediction algorithms are proposed to reduce the calculated modes by using hierarchical directional selection or spatial and temporal correlations. If all-zero blocks can be detected before DCT and quantization, the DCT and quantization can be skipped for these blocks. The computation complexity can be reduced further. Early detection algorithms of all-zero blocks for H.264 video coding have been developed in inter motion search. However, these methods cannot be used in intra prediction directly. In this paper, a novel all-zero block detection algorithm for H.264 intra prediction is proposed. In intra 4x4 modes, the SAD calculation is the same as that in inter motion search, but it is much complex in intra 16x16 modes. The Hadamard transform is used for SAD calculation of 16 AC-coefficient blocks, and an extra Hadamard transform is applied for DC block derived from them. The DC and AC thresholds are derived separately according to SAD calculations. The simulation results of all I-frame coding show that the proposed method can save up to 40% computation time in intra prediction with nearly no PSNR loss and less bitrate gain. Furthermore, it is less performance loss in the condition of I-frame and P-frame coding with the same time saving of intra prediction. Our proposed algorithm has no conflict with other fast intra prediction algorithms and can be applied on any other fast intra prediction method to achieve extra computational saving.

Paper Details

Date Published: 24 June 2005
PDF: 9 pages
Proc. SPIE 5960, Visual Communications and Image Processing 2005, 596062 (24 June 2005); doi: 10.1117/12.633386
Show Author Affiliations
Jianning Zhang, Tsinghua Univ. (China)
Lifeng Sun, Tsinghua Univ. (China)
Yuzhuo Zhong, Tsinghua Univ. (China)

Published in SPIE Proceedings Vol. 5960:
Visual Communications and Image Processing 2005
Shipeng Li; Fernando Pereira; Heung-Yeung Shum; Andrew G. Tescher, Editor(s)

© SPIE. Terms of Use
Back to Top