DocumentCode :
2526396
Title :
Can Mean Shift Trackers Perform Better?
Author :
Zhou, Huiyu ; Schaefer, Gerald ; Yuan, Yuan ; Celebi, M. Emre
Author_Institution :
Inst. of Electron., Commun. & Inf. Technol., Queen´´s Univ. Belfast, Belfast, UK
fYear :
2010
fDate :
15-18 Dec. 2010
Firstpage :
98
Lastpage :
101
Abstract :
Many tracking algorithms have difficulties dealing with occlusions and background clutters, and consequently don´t converge to an appropriate solution. Tracking based on the mean shift algorithm has shown robust performance in many circumstances but still fails e.g. when encountering dramatic intensity or colour changes in a pre-defined neighbour hood. In this paper, we present a robust tracking algorithm that integrates the advantages of mean shift tracking with those of tracking local invariant features. These features are integrated into the mean shift formulation so that tracking is performed based both on mean shift and feature probability distributions, coupled with an expectation maximisation scheme. Experimental results show robust tracking performance on a series of complicated real image sequences.
Keywords :
computer vision; expectation-maximisation algorithm; image sequences; object tracking; probability; complicated real image sequences; computer vision; expectation maximisation scheme; feature probability distributions; local invariant feature tracking; mean shift tracking algorithm; visual object tracking; Feature extraction; Image color analysis; Kernel; Probability density function; Robustness; Target tracking; Object tracking; invariants; mean shift;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Signal-Image Technology and Internet-Based Systems (SITIS), 2010 Sixth International Conference on
Conference_Location :
Kuala Lumpur
Print_ISBN :
978-1-4244-9527-6
Electronic_ISBN :
978-0-7695-4319-2
Type :
conf
DOI :
10.1109/SITIS.2010.26
Filename :
5714536
Link To Document :
بازگشت