Abstract
A new class ofGibbsian models with potentials associated with the connected components or homogeneous parts of images is introduced. For these models the neighbourhood ofa pixel is not fixed as for Markov random fields, but is given by the components which are adjacent to the pixel. The relationship to Markov random fields and marked point processes is explored and spatial Markov properties are established. Extensions to infinite lattices are also studied, and statistical inference problems including geostatistical applications and statistical image analysis are discussed. Finally, simulation studies are presented which show that the models may be appropriate for a variety of interesting patterns, including images exhibiting intermediate degrees ofspatial continuity and images of objects against background.
Original language | English |
---|---|
Journal | Advances in Applied Probability |
Volume | 30 |
Issue number | 1 |
Pages (from-to) | 1-35 |
Number of pages | 35 |
ISSN | 0001-8678 |
DOIs | |
Publication status | Published - 1 Jan 1998 |
Keywords
- Bayesian image analysis
- Exponential family models
- Geostatistics
- Infinite lattice processes
- Markov chain Monte Carlo
- Markov partitions
- Markov random fields
- Maximum likelihood
- Nearest-neighbour Markov point processes