Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/51137
Title: Optimization techniques for H.264/AVC and multi-view video
Authors: Zeng, Huanqiang
Keywords: DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
Issue Date: 2013
Abstract: The objective of this thesis is to develop optimization techniques for H.264/AVC and multi-view video. Five contributions are made in this thesis and summarized in two parts as follows. In Part I, two optimization techniques for the state-of-the-art video coding standard H.264/AVC are presented. First, an efficient intra mode decision algorithm for intra prediction in H.264/AVC, called the hierarchical intra mode decision (HIMD), is proposed. In this approach, the candidate modes are selected according to their Hadamard distances and prediction directions. Furthermore, an early-termination decision scheme with adaptive thresholding is incorporated to the proposed algorithm for further speeding up the intra mode decision process. Second, an efficient inter mode decision algorithm for inter prediction in H.264/AVC, called the motion activity-based mode decision (MAMD), is proposed. In this approach, for each macroblock (MB), only those more likely modes are chosen to be checked based on its motion activity, which can be quantitatively measured by the maximum city-block length of the motion vectors taken from a set of adjacent MBs. Experimental results have shown that both the proposed HIMD and MAMD algorithms can significantly reduce the computational complexity, while maintaining almost the same coding efficiency, compared with that of the exhaustive mode decision in the H.264/AVC.
URI: http://hdl.handle.net/10356/51137
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:EEE Theses

Files in This Item:
File Description SizeFormat 
TeG0801928G.pdf
  Restricted Access
3.54 MBAdobe PDFView/Open

Google ScholarTM

Check

Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.