Fundamentals of digital image processing
Fundamentals of digital image processing
Unsupervised Learning of Finite Mixture Models
IEEE Transactions on Pattern Analysis and Machine Intelligence
Probability Density Decomposition for Conditionally Dependent Random Variables Modeled by Vines
Annals of Mathematics and Artificial Intelligence
Robust mixture modelling using the t distribution
Statistics and Computing
Assessing a Mixture Model for Clustering with the Integrated Completed Likelihood
IEEE Transactions on Pattern Analysis and Machine Intelligence
Yet Another Survey on Image Segmentation: Region and Boundary Information Integration
ECCV '02 Proceedings of the 7th European Conference on Computer Vision-Part III
Comparing clusterings: an axiomatic view
ICML '05 Proceedings of the 22nd international conference on Machine learning
An Introduction to Copulas (Springer Series in Statistics)
An Introduction to Copulas (Springer Series in Statistics)
Toward Objective Evaluation of Image Segmentation Algorithms
IEEE Transactions on Pattern Analysis and Machine Intelligence
Unsupervised segmentation of natural images via lossy data compression
Computer Vision and Image Understanding
On the simplified pair-copula construction - Simply useful or too simplistic?
Journal of Multivariate Analysis
Online heterogeneous mixture modeling with marginal and copula selection
Proceedings of the 17th ACM SIGKDD international conference on Knowledge discovery and data mining
The infinite Student's t-mixture for robust modeling
Signal Processing
Hi-index | 0.00 |
This paper presents a finite mixture model that involves a pair-copula based construction of a multivariate distribution. The advantage of such a model is that the margins and the dependence structures are de-coupled from each other. Also, they could be modeled separately. In effect the mixture model (called DVMM) is capable of capturing a broader family of distributions including non-Gaussian models. As an application, we consider the task of color image segmentation in CIE-LUV color space. The process of segmentation could be viewed as an unsupervised clustering of the image pixels. The clusters usually represent the possible segments in the image. Here, the image pixels are assumed to be samples from a DVMM. The expectation maximization algorithm is used to estimate the model parameters. One cluster generally consists of several connected components inside an image. We further note the existence of redundant connected components that represent the boundary of two adjacent components. We here propose a methodology that merges such component pixels to the other two components. We conduct extensive experiments on Berkeley segmentation data set. We take a number of error measures to evaluate the quality of segmentation. On the basis of a comparison with two existing mixture model based segmentation approaches, we find our results encouraging.