Image Modeling and Segmentation Using Incremental Bayesian Mixture Models
Many image modeling and segmentation problems have been tackled using Gaussian Mixture Models (GMM). The two most important issues in image modeling using GMMs is the selection of the appropriate low level features and the specification of the appropriate number of GMM components. In this work we deal with the second issue and present an approach for GMM-based image modeling employing an incremental variational algorithm for Bayesian GMM training that automatically specifies the number of mixture components. Experimental results on natural and texture images indicate that the method yields reasonable models without requiring the a priori specification of the number of components.
Unable to display preview. Download preview PDF.
- 2.Li, S.Z.: Markov Random Field Modelling in Computer Vision. Springer, Heidelberg (2001)Google Scholar
- 4.Corduneanu, A., Bishop, C.M.: Variational Bayesian model selection for mixture distributions. In: Artificial Intelligence and Statistics 2001, pp. 27–34. Morgan Kaufmann, San Francisco (2001)Google Scholar
- 5.Martin, D., Fowlkes, C., Tal, D., Malik, J.: A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics. In: Proc. 8th Int’l Conf. Computer Vision, vol. 2, pp. 416–423 (2001)Google Scholar