Unsupervised Segmentation of Natural Images via Lossy Data Compression

Allen Y. Yang, John Wright, S. Shankar Sastry and Yi Ma

EECS Department
University of California, Berkeley
Technical Report No. UCB/EECS-2006-195
December 28, 2006

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2006/EECS-2006-195.pdf

In this paper, we cast natural-image segmentation as a problem of clustering texure features as multivariate mixed data. We model the distribution of the texture features using a mixture of Gaussian distributions. However, unlike most existing clustering methods, we allow the mixture components to be degenerate or nearly-degenerate. We contend that this assumption is particularly important for mid-level image segmentation, where degeneracy is typically introduced by using a common feature representation for different textures. We show that such a mixture distribution can be effectively segmented by a simple agglomerative clustering algorithm derived from a lossy data compression approach. Using simple fixed-size Gaussian windows as texture features, the algorithm segments an image by minimizing the overall coding length of all the feature vectors. In terms of a variety of performance indices, our algorithm compares favorably against other well-known image segmentation methods on the Berkeley image database.


BibTeX citation:

@techreport{Yang:EECS-2006-195,
    Author = {Yang, Allen Y. and Wright, John and Sastry, S. Shankar and Ma, Yi},
    Title = {Unsupervised Segmentation of Natural Images via Lossy Data Compression},
    Institution = {EECS Department, University of California, Berkeley},
    Year = {2006},
    Month = {Dec},
    URL = {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2006/EECS-2006-195.html},
    Number = {UCB/EECS-2006-195},
    Abstract = {In this paper, we cast natural-image segmentation as a problem of clustering texure features as multivariate mixed data. We model the distribution of the texture features using a mixture of Gaussian distributions. However, unlike most existing clustering methods, we allow the mixture components to be degenerate or nearly-degenerate. We contend that this assumption is particularly important for mid-level image segmentation, where degeneracy is typically introduced by using a common feature representation for different textures. We show that such a mixture distribution can be effectively segmented by a simple agglomerative clustering algorithm derived from a lossy data compression approach. Using simple fixed-size Gaussian windows as texture features, the algorithm segments an image by minimizing the overall coding length of all the feature vectors. In terms of a variety of performance indices, our algorithm compares favorably against other well-known image segmentation methods on the Berkeley image database.}
}

EndNote citation:

%0 Report
%A Yang, Allen Y.
%A Wright, John
%A Sastry, S. Shankar
%A Ma, Yi
%T Unsupervised Segmentation of Natural Images via Lossy Data Compression
%I EECS Department, University of California, Berkeley
%D 2006
%8 December 28
%@ UCB/EECS-2006-195
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2006/EECS-2006-195.html
%F Yang:EECS-2006-195