Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/11519
Title: Effective star-shaped search window and star-shaped pattern for motion estimation
Authors: Kam, YH
Siu, WC 
Issue Date: 2007
Source: Fourth International Conference on Information Technology and Applications, ICITA 2007, 2007, p. 290-293 How to cite?
Abstract: In previous motion estimation schemes, a search window of square shape was used, which indirectly implies that the motion activity of video objects has a square distribution. Recently, it has been suggested that the movement of video objects in the horizontal direction appears more frequently than the movement in the vertical direction in natural video sequences; such that a rectangular search window has been proposed. However, after a careful investigation, we have found that the average motion activity of actual video objects shows itself a star-shaped distribution other than a rectangular distribution. This paper proposes respectively a new shape of the search window and its search pattern for full search motion estimation (ME) which can improve the speed of the video encoding process. The proposed search window has a star shape which can cover nearly all best matches within a small region. On the other hand, the proposed search pattern provides a star-shaped searching order which is able to quicken the block matching in the partial distortion search (PDS) process. On average, more than 3 folds speedup of the PDS process can be achieved by replacing the traditional square window and the spiral pattern with the star-shaped window and the star-shaped pattern respectively while the same video quality is essentially achieved.
Description: 4th International Conference on Information Technology and Applications, ICITA 2007, Harbin, 15-18 January 2007
URI: http://hdl.handle.net/10397/11519
ISBN: 0980326702
9780980326703
Appears in Collections:Conference Paper

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

Page view(s)

52
Last Week
1
Last month
Citations as of Jul 10, 2018

Google ScholarTM

Check


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