Unsupervised Segmentation of Natural Images via Lossy Data Compression
Yang, Allen Y.
Sastry, S. Shankar
Technical Report Identifier: EECS-2006-195
December 28, 2006
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.