An Unsupervised Segmentation Method Using Markov Random Field on Region Adjacency Graph for SAR Images

Xia, Gui-Song; He, Chu

doi:10.1109/icr.2006.343148

Cited by 24 publications

(13 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, a desirable initial over-segmentation scheme should produce a minimal number of regions with boundaries preserved. Existing over-segmentation schemes include watershed transform [21]- [23], partition mode test [24], tone-region based segmentation [25], pixons extraction [20], normalized cuts [26] and superpixel lattices [27].…”

Section: Mrf-based Image Segmentationmentioning

confidence: 99%

Multivariate Image Segmentation Using Semantic Region Growing With Adaptive Edge Penalty

Qin

Clausi

2010

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

Abstract-Multivariate image segmentation is a challenging task, influenced by large intraclass variation that reduces class distinguishability as well as increased feature space sparseness and solution space complexity that impose computational cost and degrade algorithmic robustness. To deal with these problems, a Markov random field (MRF) based multivariate segmentation algorithm called "multivariate iterative region growing using semantics" (MIRGS) is presented. In MIRGS, the impact of intraclass variation and computational cost are reduced using the MRF spatial context model incorporated with adaptive edge penalty and applied to regions. Semantic region growing starting from watershed over-segmentation and performed alternatively with segmentation gradually reduces the solution space size, which improves segmentation effectiveness. As a multivariate iterative algorithm, MIRGS is highly sensitive to initial conditions. To suppress initialization sensitivity, it employs a region-level k-means

show abstract

Section: Mrf-based Image Segmentationmentioning

confidence: 99%

Multivariate Image Segmentation Using Semantic Region Growing With Adaptive Edge Penalty

Qin

Clausi

2010

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

show abstract

“…2(c)]. For instance, Xia et al [36] used the watershed to obtain the initial regions and defined the MRF on the RAG, with multilevel logistic (MLL) model for the label field and the Gamma distribution for the likelihood function. Zhang et al [39] used the mean shift to provide the initial regions instead of the watershed.…”

Section: Introductionmentioning

confidence: 99%

“…That is to say, if some regions are merged into wrong classes, the followup process cannot modify this mistake. The other optimized approach uses initial regions to generate the joint distribution and employs the maximum a posteriori (MAP) criterion to get the result of OMRF [36], [37], which provides an iterative process to improve the irreversible merging process. However, although this approach uses the likelihood function to model different image features, its segmentation accuracy is still limited since it usually treats the nodes and edges in the RAG equally for the joint distribution of the label field.…”

Section: Introductionmentioning

confidence: 99%

Semantic Segmentation of Remote Sensing Imagery Using Object-Based Markov Random Field Model With Regional Penalties

Chen

Wang

2015

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

This paper proposes a novel object-based Markov random field model (OMRF) for semantic segmentation of remote sensing images. First, the method employs the region size and edge information to build a weighted region adjacency graph (WRAG) for capturing the complicated interactions among objects. Thereafter, aimed at modeling object interactions in the OMRF, the size and edge information are further introduced into the Gibbs joint distribution of the random field as regional penalties. Finally, the semantic segmentation is achieved through a principled probabilistic inference of the OMRF with regional penalties. The proposed method is compared with other MRFbased methods and some state-of-the-art methods. Experiments are conducted on a series of synthetic and real-world images. Segmentation results demonstrate that our method provides better performance (an accuracy improvement about 3%). Moreover, we further discuss the application of the proposed method for classification.Index Terms-Object-based Markov random field (OMRF), regional penalties, remote sensing images, semantic segmentation.

show abstract

“…There have been a great deal of efforts devoted to this problem , and a large number of image models have been established over the years. Statistical models, particularly MRF model, have been successfully used as the image models for image segmentation in [3][4][5][6][7][8][9][10]. MRF model takes advantage of both regional gray scale information and Gibbs smoothness prior information in inter-pixel spatial relations.…”

Section: Introductionmentioning

confidence: 99%

A Novel Segmentation Method for Left Ventricular from Cardiac MR Images Based on Improved Markov Random Field Model

Wang

Guo

Shi

et al. 2009

2009 2nd International Congress on Image and Signal Processing

View full text Add to dashboard Cite

In this paper, we propose a improved Markov Random Field (MRF) segmentation model, which integrates region, priori knowledge and boundary information of the image, for segmenting left ventricle (LV) boundary from cardiac MR image. The proposed model incorporates geometry shape boundary information, and improves the objective function of traditional MRF model. Furthermore, Chaotic Simulated Annealing (CSA) algorithm is introduced to solve the MRF model for the first time. Since CSA algorithm introduces chaos ergodicity mechanism, it can take advantage of Chaos Algorithm (COA) and Simulated Annealing (SA) algorithm in the search process. CSA algorithm can not only avoid the limitations of mathematical optimization methods, but also greatly enhance the speed of global optimization. Experiments on clinical cardiac MR images show that the improved MRF model has high performance on segmenting LV boundary. The evaluation results illustrate that this model is robust, accurate and efficient, especially for the weak boundary and concave region .

show abstract

An Unsupervised Segmentation Method Using Markov Random Field on Region Adjacency Graph for SAR Images

Cited by 24 publications

References 13 publications

Multivariate Image Segmentation Using Semantic Region Growing With Adaptive Edge Penalty

Multivariate Image Segmentation Using Semantic Region Growing With Adaptive Edge Penalty

Semantic Segmentation of Remote Sensing Imagery Using Object-Based Markov Random Field Model With Regional Penalties

A Novel Segmentation Method for Left Ventricular from Cardiac MR Images Based on Improved Markov Random Field Model

Contact Info

Product

Resources

About