Synthetic aperture radar image segmentation by a detail preserving Markov random field approach

Smits, P.C.; Dellepiane, S.

doi:10.1109/36.602527

Cited by 78 publications

(47 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The MRF approach has proven to be very successful in addressing the issues of both data modelling and regularization in segmentation and classification by means of a-priori information and other sources of data (e.g., digital read maps, optical data, etc.) [2], [3], [4], [5], [6]. But also filtering can be performed by using suited MRF models, as has been shown in [9] and [ 12].…”

Section: Results To Datementioning

confidence: 99%

Analysis and segmentation of remote-sensing images for land-cover mapping

Smits

Serpico

1997

Image Analysis and Processing

View full text Add to dashboard Cite

Abstract. In the framework of the European Community programme Training and Mobility for Researchers, the project Analysis and Segmentation of Remote-Sensing Images for Land-Cover mapping has been proposed and approved. This article provides some insight in the role of pattern recognition and image processing techniques in the European remote-sensing community and gives and overview of the project's objectives and results to date. Remote-sensing and image analysisRemote Sensing exists by the virtue of physical, economical, political or other constraints that prevent human beings from exploring certain regions. Although many sophisticated instruments have been developed, the hunger for information seems more than ever unsatisfyable, and the more types of information can be extracted from new or existing remote-sensing images, the more economical attractive these data become.It is not our intention to go too much into detail on all the aspects that are related to remote-sensing image analysis. We will limit ourselves to some observations that will help to understand the scientific direction of the project Analysis and Segmentation of Remote-Sensing Images for Land-Cover mapping (ANRS).For an up-to-date, complete and comprehensive textbook on remote-sensing image analysis the reader is referred to [1].Segmentation, classification, and change-detection of remote-sensing images can be formulated as ill-posed problems. Although the quality of many modern sensors is such that these problems are not too ill-posed, this situation changes in cases where one wants to extract types of information for which the sensor has not been build in the first place. This can be the case, for example, when one is interested in the recognition of objects with dimensions close to the pixel resolution.When dealing with complex natural and agricultural landscapes, characterized by the presence of mixed classes and by land-cover/ground responses influenced by terrain topographies, it is not sufficient to base decisions regarding a data element only on the single element itself coming from a single sensor. Regularization of the ill-posed analysis problem is needed such that the uncertainty is reduced sharply, thus improving the consistency of the solution. The information needed to regularize a problem can come, for instance, from

show abstract

Section: Results To Datementioning

confidence: 99%

Analysis and segmentation of remote-sensing images for land-cover mapping

Smits

Serpico

1997

Image Analysis and Processing

View full text Add to dashboard Cite

show abstract

“…48 , and at each site we select the one that leads to the smallest energy 48,49 . The adaptivity of the neighborhood aims to take into account the geometrical properties of the different areas in our original image, a spatial feature that plays a primary role especially in high-resolution imagery.…”

Section: Prior Probabilitymentioning

confidence: 99%

Multichannel hierarchical image classification using multivariate copulas

et al. 2012

View full text Add to dashboard Cite

International audienceThis paper focuses on the classification of multichannel images. The proposed supervised Bayesian classification method applied to histological (medical) optical images and to remote sensing (optical and synthetic aperture radar) imagery consists of two steps. The first step introduces the joint statistical modeling of the coregistered input images. For each class and each input channel, the class-conditional marginal probability density functions are estimated by finite mixtures of well-chosen parametric families. For optical imagery, the normal distribution is a well-known model. For radar imagery, we have selected generalized gamma, log-normal, Nakagami and Weibull distributions. Next, the multivariate d-dimensional Clayton copula, where d can be interpreted as the number of input channels, is applied to estimate multivariate joint class-conditional statistics. As a second step, we plug the estimated joint probability density functions into a hierarchical Markovian model based on a quad-tree structure. Multiscale features are extracted by discrete wavelet transforms, or by using input multiresolution data. To obtain the classification map, we integrate an exact estimator of the marginal posterior mode

show abstract

“…In this experiment, the boundary between two regions was extracted using the simple Sobel operator. More sophisticated model-based methods of an automatic image segmentation, e.g., methods such as proposed in [26]- [28], would potentially perform better than the Sobel operator.…”

Section: B Application To a Real Imagementioning

confidence: 99%

Comparison of GLR and invariant detectors under structured clutter covariance

Kim

Hero

2001

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

Abstract-This paper addresses a target detection problem in radar imaging for which the covariance matrix of unknown Gaussian clutter has block diagonal structure. This block diagonal structure is the consequence of a target lying along a boundary between two statistically independent clutter regions. Here, we design adaptive detection algorithms using both the generalized likelihood ratio (GLR) and the invariance principles. There has been considerable recent interest in applying invariant hypothesis testing as an alternative to the GLR test. This interest has been motivated by several attractive properties of invariant tests including: exact robustness to variation of nuisance parameters and possible finite-sample min-max optimality. However, in our deep-hide target detection problem, there are regimes for which neither the GLR nor the invariant tests uniformly outperforms the other. We will discuss the relative advantages of GLR and invariance procedures in the context of this radar imaging and target detection application.

show abstract

Synthetic aperture radar image segmentation by a detail preserving Markov random field approach

Cited by 78 publications

References 26 publications

Analysis and segmentation of remote-sensing images for land-cover mapping

Analysis and segmentation of remote-sensing images for land-cover mapping

Multichannel hierarchical image classification using multivariate copulas

Comparison of GLR and invariant detectors under structured clutter covariance

Contact Info

Product

Resources

About