DocumentCode :
41428
Title :
A Nonsymmetric Mixture Model for Unsupervised Image Segmentation
Author :
Thanh Minh Nguyen ; Wu, Q. M. Jonathan
Author_Institution :
Dept. of Electr. & Comput. Eng., Univ. of Windsor, Windsor, ON, Canada
Volume :
43
Issue :
2
fYear :
2013
fDate :
Apr-13
Firstpage :
751
Lastpage :
765
Abstract :
Finite mixture models with symmetric distribution have been widely used for many computer vision and pattern recognition problems. However, in many applications, the distribution of the data has a non-Gaussian and nonsymmetric form. This paper presents a new nonsymmetric mixture model for image segmentation. The advantage of our method is that it is simple, easy to implement, and intuitively appealing. In this paper, each label is modeled with multiple D-dimensional Student´s t-distribution, which is heavily tailed and more robust than Gaussian distribution. Expectation-maximization algorithm is adopted to estimate model parameters and to maximize the lower bound on the data log-likelihood from observations. Numerical experiments on various data types are conducted. The performance of the proposed model is compared with that of other mixture models, demonstrating the robustness, accuracy, and effectiveness of our method.
Keywords :
expectation-maximisation algorithm; image resolution; image segmentation; parameter estimation; statistical distributions; unsupervised learning; computer vision problem; data distribution; expectation-maximization algorithm; finite mixture models; image pixels; model parameter estimation; multiple D-dimensional student t-distribution; nonGaussian form; nonsymmetric form; nonsymmetric mixture model; pattern recognition problem; symmetric distribution; unsupervised image segmentation; Data models; Gaussian distribution; Image segmentation; Numerical models; Parameter estimation; Robustness; Shape; Expectation–maximization (EM) algorithm; non-Gaussian distribution; nonsymmetric mixture model (NSMM); unsupervised image segmentation;
fLanguage :
English
Journal_Title :
Cybernetics, IEEE Transactions on
Publisher :
ieee
ISSN :
2168-2267
Type :
jour
DOI :
10.1109/TSMCB.2012.2215849
Filename :
6298972
Link To Document :
بازگشت