DocumentCode :
1741046
Title :
Impulse restoration template matching under geometric uncertainties
Author :
Dufour, Roger M. ; Miller, Eric L. ; Galatsanos, Nikolas R.
Author_Institution :
Northeastern Univ., Boston, MA, USA
Volume :
2
fYear :
2000
fDate :
10-13 Sept. 2000
Firstpage :
950
Abstract :
A common image processing problem is determining the location of an object using a template when the size and rotation of the target are unknown. Using a maximum likelihood formulation, localization is possible by computing a likelihood surface for a dense sampling of the size and rotation space. However, optimization is complicated by local minima and regions of small or zero gradient. We therefore demonstrate a technique which employs a library of templates starting from a smooth approximation and adding detail until the exact template is reached. Successively estimating the geometric parameters using these templates achieves the accuracy of the exact template while remaining within a well-behaved "bowl" in the search space which allows standard minimization techniques to be used.
Keywords :
Newton method; image matching; image restoration; image sampling; maximum likelihood estimation; optimisation; Newton algorithm; dense sampling; geometric parameters estimation; geometric uncertainties; image processing; impulse restoration template matching; likelihood surface; local minima; maximum likelihood formulation; minimization techniques; object location determination; optimization; rotation space; search space; smooth approximation; target rotation; target size; templates library; zero gradient image regions; Equations; Image processing; Image restoration; Image sampling; Libraries; Matched filters; Parameter estimation; Space technology; Tail; Uncertainty;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Image Processing, 2000. Proceedings. 2000 International Conference on
Conference_Location :
Vancouver, BC, Canada
ISSN :
1522-4880
Print_ISBN :
0-7803-6297-7
Type :
conf
DOI :
10.1109/ICIP.2000.899874
Filename :
899874
Link To Document :
بازگشت