scispace - formally typeset
Search or ask a question
Topic

Markov random field

About: Markov random field is a research topic. Over the lifetime, 5669 publications have been published within this topic receiving 179568 citations. The topic is also known as: MRF.


Papers
More filters
Journal ArticleDOI
TL;DR: This paper develops a new method to estimate the parameters defining the Markovian distribution of the measured data, while performing the data clustering simultaneously, using fuzzy Markov random fields (fuzzy MRFs) for the segmentation of multispectral MR prostate images.
Abstract: Prostate cancer is one of the leading causes of death from cancer among men in the United States. Currently, high-resolution magnetic resonance imaging (MRI) has been shown to have higher accuracy than trans-rectal ultrasound (TRUS) when used to ascertain the presence of prostate cancer. As MRI can provide both morphological and functional images for a tissue of interest, some researchers are exploring the uses of multispectral MRI to guide prostate biopsies and radiation therapy. However, success with prostate cancer localization based on current imaging methods has been limited due to overlap in feature space of benign and malignant tissues using any one MRI method and the interobserver variability. In this paper, we present a new unsupervised segmentation method for prostate cancer detection, using fuzzy Markov random fields (fuzzy MRFs) for the segmentation of multispectral MR prostate images. Typically, both hard and fuzzy MRF models have two groups of parameters to be estimated: the MRF parameters and class parameters for each pixel in the image. To date, these two parameters have been treated separately, and estimated in an alternating fashion. In this paper, we develop a new method to estimate the parameters defining the Markovian distribution of the measured data, while performing the data clustering simultaneously. We perform computer simulations on synthetic test images and multispectral MR prostate datasets to demonstrate the efficacy and efficiency of the proposed method and also provide a comparison with some of the commonly used methods.

136 citations

Proceedings ArticleDOI
11 Feb 2019
TL;DR: This paper showed that BERT is a Markov random field language model and showed that it can produce high quality, fluent generations, compared to the generations of a traditional left-to-right language model.
Abstract: We show that BERT (Devlin et al., 2018) is a Markov random field language model. This formulation gives way to a natural procedure to sample sentences from BERT. We generate from BERT and find that it can produce high quality, fluent generations. Compared to the generations of a traditional left-to-right language model, BERT generates sentences that are more diverse but of slightly worse quality.

136 citations

Journal ArticleDOI
TL;DR: The experiments that were performed on a bitemporal TerraSAR-X StripMap data set from South West England during and after a large-scale flooding in 2007 confirm the effectiveness of the proposed change detection method and show an increased classification accuracy of the hybrid MRF model in comparison to the sole application of the HMAP estimation.
Abstract: The near real-time provision of precise information about flood dynamics from synthetic aperture radar (SAR) data is an essential task in disaster management. A novel tile-based parametric thresholding approach under the generalized Gaussian assumption is applied on normalized change index data to automatically solve the three-class change detection problem in large-size images with small class a priori probabilities. The thresholding result is used for the initialization of a hybrid Markov model which integrates scale-dependent and spatiocontextual information into the labeling process by combining hierarchical with noncausal Markov image modeling. Hierarchical maximum a posteriori (HMAP) estimation using the Markov chains in scale, originally developed on quadtrees, is adapted to hierarchical irregular graphs. To reduce the computational effort of the iterative optimization process that is related to noncausal Markov models, a Markov random field (MRF) approach is defined, which is applied on a restricted region of the lowest level of the graph, selected according to the HMAP labeling result. The experiments that were performed on a bitemporal TerraSAR-X StripMap data set from South West England during and after a large-scale flooding in 2007 confirm the effectiveness of the proposed change detection method and show an increased classification accuracy of the hybrid MRF model in comparison to the sole application of the HMAP estimation. Additionally, the impact of the graph structure and the chosen model parameters on the labeling result as well as on the performance is discussed.

136 citations

Journal ArticleDOI
TL;DR: It is shown that dual polarization SAR data can yield segmentation resultS similar to those obtained with fully polarimetric SAR data, and the performance of the MAP segmentation technique is evaluated.
Abstract: A statistical image model is proposed for segmenting polarimetric synthetic aperture radar (SAR) data into regions of homogeneous and similar polarimetric backscatter characteristics. A model for the conditional distribution of the polarimetric complex data is combined with a Markov random field representation for the distribution of the region labels to obtain the posterior distribution. Optimal region labeling of the data is then defined as maximizing the posterior distribution of the region labels given the polarimetric SAR complex data (maximum a posteriori (MAP) estimate). Two procedures for selecting the characteristics of the regions are then discussed. Results using real multilook polarimetric SAR complex data are given to illustrate the potential of the two selection procedures and evaluate the performance of the MAP segmentation technique. It is also shown that dual polarization SAR data can yield segmentation resultS similar to those obtained with fully polarimetric SAR data. >

135 citations

Journal ArticleDOI
TL;DR: An evolutionary approach, selectionist relaxation, is proposed as a solution to the problem of segmenting Markov random field modeled textures in unsupervised mode and the generalized Ising model is used to represent textured data.
Abstract: Among the existing texture segmentation methods, those relying on Markov random fields have retained substantial interest and have proved to be very efficient in supervised mode. The use of Markov random fields in unsupervised mode is, however, hampered by the parameter estimation problem. The recent solutions proposed to overcome this difficulty rely on assumptions about the shapes of the textured regions or about the number of textures in the input image that may not be satisfied in practice. In this paper, an evolutionary approach, selectionist relaxation, is proposed as a solution to the problem of segmenting Markov random field modeled textures in unsupervised mode. In selectionist relaxation, the computation is distributed among a population of units that iteratively evolves according to simple and local evolutionary rules. A unit is an association between a label and a texture parameter vector. The units whose likelihood is high are allowed to spread over the image and to replace the units that receive lower support from the data. Consequently, some labels are growing while others are eliminated. Starting with an initial random population, this evolutionary process eventually results in a stable labelization of the image, which is taken as the segmentation. In this work, the generalized Ising model is used to represent textured data. Because of the awkward nature of the partition function in this model, a high-temperature approximation is introduced to allow the evaluation of unit likelihoods. Experimental results on images containing various synthetic and natural textures are reported.

135 citations


Network Information
Related Topics (5)
Image segmentation
79.6K papers, 1.8M citations
94% related
Convolutional neural network
74.7K papers, 2M citations
93% related
Feature extraction
111.8K papers, 2.1M citations
92% related
Image processing
229.9K papers, 3.5M citations
91% related
Deep learning
79.8K papers, 2.1M citations
91% related
Performance
Metrics
No. of papers in the topic in previous years
YearPapers
20241
202330
2022128
202196
2020173
2019204