Image Segmentation Using Redundancy Reduction
In this research, we cast natural-image segmentation as a problem of clustering feature vectors extracted from image textures. We model the distribution of the texture features using a mixture of von Mises-Fisher distributions. In the proposed algorithm, by fitting these models to the data we segment the im- age using an agglomerative clustering algorithm based on the likelihood of the fit. We conduct comprehensive experiments to measure the performance of the algorithm in terms of visual evaluation and a variety of quantitative indices, and we also compare it to other well-known image-segmentation methods. While moving the project forward, we have also developed a software pack- age for working with mixture models, which provides the facilities required to estimate the parameters of these models.
Mixest: An estimation toolbox for mixture models
arXiv preprint arXiv:1507.06065, 2015
K-means++ for Mixtures of von Mises-Fisher Distributions
IKT2015 7th International Conference on Information and Knowledge Technology, 2015