@inproceedings {238, title = {Automated classification of operational SAR sea ice images}, booktitle = {7th Canadian Conference on Computer and Robot Vision}, year = {2010}, month = {06/2010}, address = {Ottawa, Ontario, Canada}, abstract = {

The automated classification of operational sea ice satellite imagery is important for ship navigation and environmental monitoring. Annually, thousands of large synthetic aperture radar (SAR) scenes are manually processed by the Canadian Ice Service (CIS) and pixel-level interpretation is not feasible. Trained ice analysts divide SAR images into $\#$x201D;polygon $\#$x201D; areas and then identify the number and type of ice classes per polygon. Full scene unsupervised classification can be performed by first segmenting each polygon into distinct regions algorithmically. Since there is insufficient information to assign a sea ice label for each region within an individual polygon, a Markov random field formulation using joint information to label each region in a full SAR scene has been developed. This approach has been successfully applied to operational CIS data to produce pixel-level classified images and is the first known successful end-to-end process for automatically classifying operational SAR sea ice images.

}, keywords = {automated classification, Canadian Ice Service, environmental monitoring, geophysical image processing, image segmentation, Markov processes, Markov random field, navigation, operational SAR sea ice images, pattern classification, pixel level interpretation, sea ice, sea ice satellite imagery, ship navigation, synthetic aperture radar, synthetic aperture radar scenes, unsupervised classification}, doi = {http://dx.doi.org/10.1109/CRV.2010.59}, author = {S Ochilov and D A. Clausi} } @article {225, title = {Multivariate image segmentation using semantic region growing with adaptive edge penalty}, journal = {IEEE Transactions on Image Processing}, volume = {19}, year = {2010}, pages = {2157 - 2170}, 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 $\#$x201C;multivariate iterative region growing using semantics $\#$x201D; (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 (RKM) based initialization method, which consistently provides accurate initial conditions at low computational cost. Experiments show the superiority of RKM relative to two commonly used initialization methods. Segmentation tests on a variety of synthetic and natural multivariate images demonstrate that MIRGS consistently outperforms three other published algorithms.

}, keywords = {adaptive edge penalty, Algorithms, Automated, Computer-Assisted, image enhancement, image interpretation, image segmentation, iterative methods, Markov processes, Markov random field, MRF spatial context model, Multivariate Analysis, multivariate image segmentation, multivariate iterative region growing, Pattern Recognition, region-level k -means based initialization, Reproducibility of Results, semantic region growing, Semantics, Sensitivity and Specificity}, issn = {1057-7149}, doi = {http://dx.doi.org/10.1109/TIP.2010.2045708}, author = {K Qin and D A. Clausi} } @article {383, title = {Filament preserving model (FPM) segmentation applied to SAR sea-ice imagery}, journal = {IEEE Transactions on Geoscience and Remote Sensing}, volume = {44}, year = {2006}, pages = {3687 - 3694}, abstract = {

Modeling spatial context constraints using a Markov random field (MRF) has been widely used in the segmentation of noisy images. Its applicability to synthetic aperture radar (SAR) sea-ice segmentation has also been demonstrated recently. However, most existing MRF models are not capable of preserving filaments, specifically leads and ridges for SAR sea ice, which are valuable for ship navigation applications and necessary for identifying certain ice types. In this paper, a new statistical context model is proposed that, within the same scene, can simultaneously preserve narrow elongated features while producing similar smooth segmentation results comparable to typical MRF-based approaches. Tested on one synthetic image and two SAR sea-ice scenes, this filament preserving model substantially improves classification accuracies when compared to standard Gaussian mixture and MRF-based segmentation algorithms

}, keywords = {egg code, filament preserving model, image classification, image segmentation, Markov processes, Markov random field, MRF-based segmentation algorithm, noisy image segmentation, oceanographic techniques, remote sensing by radar, SAR sea-ice imagery, sea ice, ship navigation, standard Gaussian mixture, synthetic aperture radar, synthetic image}, issn = {0196-2892}, doi = {http://dx.doi.org/10.1109/TGRS.2006.885046}, author = {Q Yu and D A. Clausi} } @inproceedings {393, title = {Filament preserving segmentation for SAR sea ice imagery using a new statistical model}, booktitle = {18th International Conference on Pattern Recognition (ICPR)}, year = {2006}, month = {08/2006}, address = {Hong Kong}, abstract = {

Modelling spatial context constraints using Markov random field (MRF) has been widely used in the segmentation of noisy images. Its applicability to SAR sea ice segmentation has also been demonstrated by Deng and Clausi (2005). However, most existing MRF models are not capable of preserving filaments, specifically leads and ridges for SAR sea ice, which are valuable for ship navigation applications and helpful for identifying certain ice types. A new statistical context model is proposed that can preserve such narrow elongated features while producing similar smooth segmentation results as those of existing MRF based approaches

}, keywords = {Feature Extraction, filament preserving segmentation, geophysical signal processing, Image Denoising, image segmentation, Markov processes, Markov random field, narrow elongated features, noisy image segmentation, oceanographic techniques, radar imaging, SAR sea ice imagery, sea ice, ship navigation, spatial context constraints, statistical analysis, statistical model, synthetic aperture radar}, doi = {http://dx.doi.org/10.1109/ICPR.2006.561}, author = {Q Yu and D A. Clausi} } @inproceedings {644, title = {Combining local and global features for image segmentation using iterative classification and region merging}, booktitle = {2nd Canadian Conference on Computer and Robot Vision}, year = {2005}, month = {05/2005}, abstract = {

In MRF based unsupervised segmentation, the MRF model parameters are typically estimated globally. Those global statistics sometimes are far from accurate for local areas if the image is highly non-stationary, and hence will generate false boundaries. The problem cannot be solved if local statistics are not considered. This work incorporates the local feature of edge strength in the MRF energy function, and segmentation is obtained by reducing the energy function using iterative classification and region merging.

}, keywords = {edge detection, edge strength, Feature Extraction, global feature, global statistics, image classification, image segmentation, iterative classification, iterative methods, local feature, local statistics, Markov processes, Markov random field, MRF based unsupervised segmentation, MRF energy function, MRF model parameter, region merging}, author = {Q Yu and D A. Clausi} } @inproceedings {427, title = {Image denoising using complex wavelets and Markov prior models}, booktitle = {12th IEEE International Conference on Image Processing}, year = {2005}, publisher = {Springer}, organization = {Springer}, abstract = {

We combine the techniques of the complex wavelet transform and Markov random fields (MRF) model to restore natural images in white Gaussian noise. The complex wavelet transform outperforms the standard real wavelet transform in the sense of shift-invariance, directionality and complexity. The prior MRF model is used to exploit the clustering property of the wavelet transform, which can effectively remove annoying pointlike artifacts associated with standard wavelet denoising methods. Our experimental results significantly outperform those using standard wavelet transforms and are comparable to those from overcomplete wavelet transforms and MRFs, but with much less complexity.

}, keywords = {complex wavelet transform, Image Denoising, Markov random field}, doi = {http://dx.doi.org/10.1007/11559573_10}, author = {F Jin and P Fieguth and L Winger} } @article {417, title = {Operational map-guided classification of SAR sea ice imagery}, journal = {IEEE Transactions on Geoscience and Remote Sensing}, volume = {43}, year = {2005}, pages = {2940 - 2951}, abstract = {

This paper presents a map-guided sea ice classification system built to work in parallel with the Canadian Ice Service (CIS) operations to produce pixel-based ice maps that complement actual \"egg code\" maps produced by CIS. The system uses the CIS maps as input to guide classification by providing information on the number of ice types and their final label for specific regions. Segmentation is based on a modified adaptive Markov random field (MRF) model that uses synthetic aperture radar (SAR) intensities and texture features as input. The ice type labeling is performed automatically by gathering evidences based on a priori information on one or two classes and deducing the other labels iteratively by comparing distributions of segments. Three methods for comparing the segment distributions (Fisher criterion, Mahalanobis distance, and Kolmogorov-Smirnov test) were implemented and compared. The system is fully described with special attention to the labeling procedure. Examples are presented in the form of two CIS SAR-based ice maps from the Gulf of Saint Lawrence region and one example from the Beaufort Sea. The results indicate that when the segmentation is good, the labeling attains best results (between 71\% and 89\%) based on evaluation by a sea ice analyst. Some problems remain to be assessed which are primarily attributable to discrepancies in the information provided by the egg code and what is actually visible in the SAR image. Subscale information on floe size and shape available to human analysts, but not in this classification system, also appear to be a critical information for separating some ice types.

}, keywords = {Beaufort Sea, Canadian Ice Service, Fisher criterion, geophysical signal processing, gray-level cooccurrence matrix, Gulf of Saint Lawrence, image classification, image segmentation, image texture, Kolmogorov-Smirnov test, Mahalanobis distance, map-guided classification, Markov processes, Markov random field, oceanographic regions, oceanographic techniques, pixel-based ice maps, radar imaging, remote sensing by laser beam, remote sensing by radar, SAR sea ice imagery, sea ice, sea ice classification, segment distribution comparison, synthetic aperture radar}, issn = {0196-2892}, doi = {http://dx.doi.org/10.1109/TGRS.2005.857897}, author = {P Maillard and D A. Clausi and H Deng} } @inproceedings {694, title = {Advanced Gaussian MRF rotation-invariant texture features for classification of remote sensing imagery}, booktitle = {2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition}, year = {2003}, address = {Madison, WI, USA}, abstract = {

The features based on Markov random field (MRF) models are usually sensitive to the rotation of image textures. The paper develops an anisotropic circular Gaussian MRF (ACGMRF) model for modeling rotated image textures and retrieving rotation-invariant texture features. To overcome the singularity problem of the least squares estimate (LSE) method, an approximate least squares estimate (ALSE) method is proposed to estimate the parameters of ACGMRF model. The rotation-invariant features can be obtained from the parameters of the ACGMRF model by the one-dimensional (1D) discrete Fourier transform (DFT). Significantly improved accuracy can be achieved by applying the rotation-invariant features to classify SAR (synthetic aperture radar) sea ice and Brodatz imagery.

}, keywords = {ACGMRF model, ALSE method, anisotropic circular Gaussian MRF, approximate least squares estimate, Brodatz imagery, discrete Fourier transform, discrete Fourier transforms, Feature Extraction, Gaussian distribution, image classification, image texture rotation, least squares approximation, least squares approximations, Markov processes, Markov random field, one-dimensional DFT, Remote Sensing, remote sensing image classification, rotated image texture modeling, rotation-invariant feature, rotation-invariant texture feature, SAR classification, sea ice, singularity problem, synthetic aperture radar}, doi = {http://dx.doi.org/10.1109/CVPR.2003.1211533}, author = {H Deng and D A. Clausi} } @inproceedings {602, title = {A probabilistic framework for image segmentation}, booktitle = {IEEE International Conference on Image Processing}, year = {2003}, abstract = {

A new probabilistic image segmentation model based on hypothesis testing and Gibbs random fields is introduced. First, a probabilistic difference measure derived from a set of hypothesis tests is introduced. Next, a Gibbs/Markov random field model endowed with the new measure is then applied to the image segmentation problem to determine the segmented image directly through energy minimization. The Gibbs/Markov random fields approach permits us to construct a rigorous computational framework where local and regional constraints can be globally optimized. Results on grayscale and color images are encouraging.

}, keywords = {energy minimization, Gibbs random field, hypothesis testing, image segmentation, Markov random field, probabilistic image segmentation, probability}, doi = {http://dx.doi.org/10.1109/ICIP.2003.1246714}, author = {S Wesolkowski and P Fieguth} }