Abstract
We consider the problem of multiband image clustering and segmentation. We propose a new methodology for doing this, called model-based cluster trees. This is grounded in model-based clustering, which bases inference on finite mixture models estimated by maximum likelihood using the EM algorithm, and automatically chooses the number of clusters by Bayesian model selection, approximated using BIC, the Bayesian Information Criterion. For segmentation, model-based clustering is based on a Markov spatial dependence model. In the Markov model case, the Bayesian model selection criterion takes account of spatial neighborhood information, and is termed PLIC, the Pseudolikelihood Information Criterion. We build a cluster tree by first segmenting an image band, then using the second band to cluster each of the level 1 clusters, and continuing if required for further bands. The tree is pruned automatically as a part of the algorithm by using Bayesian model selection to choose the number of clusters at each stage. An efficient algorithm for implementing the methodology is proposed. An example is used to evaluate this new approach, and the advantages and disadvantages of alternative approaches to multiband segmentation and clustering are discussed.
| Language | English |
|---|---|
| Pages | 587-596 |
| Number of pages | 10 |
| Journal | Image and Vision Computing |
| Volume | 23 |
| Issue number | 6 |
| Early online date | 28 Mar 2005 |
| DOIs | |
| Publication status | Published - 1 Jun 2005 |
| Externally published | Yes |
Fingerprint
Cite this
}
Bayesian inference for multiband image segmentation via model-based cluster trees. / Murtagh, Fionn; Raftery, Adrian E.; Starck, Jean Luc.
In: Image and Vision Computing, Vol. 23, No. 6, 01.06.2005, p. 587-596.Research output: Contribution to journal › Article
TY - JOUR
T1 - Bayesian inference for multiband image segmentation via model-based cluster trees
AU - Murtagh, Fionn
AU - Raftery, Adrian E.
AU - Starck, Jean Luc
PY - 2005/6/1
Y1 - 2005/6/1
N2 - We consider the problem of multiband image clustering and segmentation. We propose a new methodology for doing this, called model-based cluster trees. This is grounded in model-based clustering, which bases inference on finite mixture models estimated by maximum likelihood using the EM algorithm, and automatically chooses the number of clusters by Bayesian model selection, approximated using BIC, the Bayesian Information Criterion. For segmentation, model-based clustering is based on a Markov spatial dependence model. In the Markov model case, the Bayesian model selection criterion takes account of spatial neighborhood information, and is termed PLIC, the Pseudolikelihood Information Criterion. We build a cluster tree by first segmenting an image band, then using the second band to cluster each of the level 1 clusters, and continuing if required for further bands. The tree is pruned automatically as a part of the algorithm by using Bayesian model selection to choose the number of clusters at each stage. An efficient algorithm for implementing the methodology is proposed. An example is used to evaluate this new approach, and the advantages and disadvantages of alternative approaches to multiband segmentation and clustering are discussed.
AB - We consider the problem of multiband image clustering and segmentation. We propose a new methodology for doing this, called model-based cluster trees. This is grounded in model-based clustering, which bases inference on finite mixture models estimated by maximum likelihood using the EM algorithm, and automatically chooses the number of clusters by Bayesian model selection, approximated using BIC, the Bayesian Information Criterion. For segmentation, model-based clustering is based on a Markov spatial dependence model. In the Markov model case, the Bayesian model selection criterion takes account of spatial neighborhood information, and is termed PLIC, the Pseudolikelihood Information Criterion. We build a cluster tree by first segmenting an image band, then using the second band to cluster each of the level 1 clusters, and continuing if required for further bands. The tree is pruned automatically as a part of the algorithm by using Bayesian model selection to choose the number of clusters at each stage. An efficient algorithm for implementing the methodology is proposed. An example is used to evaluate this new approach, and the advantages and disadvantages of alternative approaches to multiband segmentation and clustering are discussed.
KW - Bayesian model
KW - Clustering
KW - Hyperspectral
KW - Information criterion
KW - Information fusion
KW - Ising
KW - Markov model
KW - Multiband
KW - Multichannel
KW - Multispectral
KW - Potts
KW - Quantization
KW - Segmentation
UR - http://www.scopus.com/inward/record.url?scp=17444368072&partnerID=8YFLogxK
U2 - 10.1016/j.imavis.2005.02.002
DO - 10.1016/j.imavis.2005.02.002
M3 - Article
VL - 23
SP - 587
EP - 596
JO - Image and Vision Computing
T2 - Image and Vision Computing
JF - Image and Vision Computing
SN - 0262-8856
IS - 6
ER -