Title :
Effect of noise on model selection criteria in visual applications
Author :
Gheissari, Niloofar ; Bab-Hadiashar, Alireza
Author_Institution :
Swinburne Univ. of Technol., Melbourne, Vic., Australia
Abstract :
In this paper, the effect of noise level on the performance of a large number of existing model selection criteria for two important computer vision applications (motion and range segmentation) has been investigated. The results of our experiments show that although the performance of all model selection criteria deteriorates by increasing the level of noise, the surface selection criterion (SSC) remains the criterion of choice for different noise levels.
Keywords :
computer vision; image motion analysis; image segmentation; noise; computer vision applications; model selection criteria; motion segmentation; range segmentation; visual applications; Application software; Australia; Computer vision; H infinity control; Layout; Libraries; Noise level; Parametric statistics; Polynomials; Shape;
Conference_Titel :
Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on
Print_ISBN :
0-7695-2128-2
DOI :
10.1109/ICPR.2004.1334102
