Title :
Kernel based image registration versus MLESAC: A comparative study
Author :
Fuiorea, Daniela ; Gui, Vasile ; Pescaru, Dan ; Toma, Corneliu
Author_Institution :
Commun. Dept., Politeh. Univ. of Timisoara, Timisoara, Romania
Abstract :
This paper evaluates the performance of a nonparametric robust image registration method based on the mean shift algorithm, which could successfully replace the random sampling algorithms. Therefore it realizes a comparative study between the proposed nonparametric method and other two important robust random sampling methods, RANSAC and MLESAC. These techniques are analyzed and tested for performance evaluation in several image registration scenarios.
Keywords :
image registration; image sampling; maximum likelihood estimation; MLESAC; kernel based image registration method; maximum likelihood estimation sample consensus; mean shift algorithm; performance evaluation; random sampling method; Cost function; Image registration; Image sampling; Iterative algorithms; Kernel; Maximum likelihood estimation; Noise robustness; Parameter estimation; Probability density function; Solid modeling;
Conference_Titel :
Applied Computational Intelligence and Informatics, 2009. SACI '09. 5th International Symposium on
Conference_Location :
Timisoara
Print_ISBN :
978-1-4244-4477-9
Electronic_ISBN :
978-1-4244-4478-6
DOI :
10.1109/SACI.2009.5136252
