Please use this identifier to cite or link to this item:
Title: Yet a faster motion estimation algorithm with directional search strategies
Authors: Zhang, Y
Siu, WC 
Shen, T
Issue Date: 2007
Source: 2007 15th International Conference on Digital Signal Processing, DSP 2007, 2007, 4288622, p. 475-478
Abstract: An efficient fast search motion estimation algorithm is highly desired in many video coding applications. Most previous fast algorithms are performed in a blind way, without critically making use of the characteristics of motion information of the sequence being coded. In this paper, we propose a new fast search motion estimation algorithm, which uses the directional information of the selected checking points obtained in the search procedure to guide the search. The statistical information which features the motion activities of the blocks in the previous frame is used to predict the characteristics of the motion activities of the blocks in the current frame. This is to save the computational cost and to avoid spending effort on the blocks which will unlikely provide the optimal match. Hence, the computational resource can be re-assigned to locations that deserve to be searched more than others. Extensive experimental work has been done, results of which show that our approach gives a speedup of 5.39 to 1.14 times over that of the recent fast algorithms and 150 times over the exhaustive full search algorithm on average, with a negligible degradation in peak signalto-noise ratio (PSNR).
Keywords: Directional search
Early termination
Motion estimation
Motion vector
Video coding
ISBN: 1424408822
DOI: 10.1109/ICDSP.2007.4288622
Description: 2007 15th International Conference onDigital Signal Processing, DSP 2007, Wales, 1-4 July 2007
Appears in Collections:Conference Paper

View full-text via PolyU eLinks SFX Query
Show full item record


Last Week
Last month
Citations as of Sep 7, 2020

Page view(s)

Last Week
Last month
Citations as of Sep 13, 2020

Google ScholarTM



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.