Show simple item record

dc.contributor.authorXia, Gui-Song
dc.contributor.authorTriggs, Bill
HAL ID: 741773
ORCID: 0000-0003-4116-6296
dc.contributor.authorDai, Dengxin
dc.contributor.authorYang, Wen
dc.subjectsynthetic aperture radaren
dc.subjectscene labelingen
dc.subjectprobabilistic latent semantic analysis (PLSA)en
dc.subjectHierarchical Markov aspect model (HMAM)en
dc.titleSAR-based Terrain Classification using Weakly Supervised Hierarchical Markov Aspect Modelsen
dc.typeArticle accepté pour publication ou publié
dc.contributor.editoruniversityotherLaboratoire Jean Kuntzmann (LJK) CNRS : UMR5224 – Université Joseph Fourier - Grenoble I – Université Pierre-Mendès-France - Grenoble II – Institut Polytechnique de Grenoble - Grenoble Institute of Technology;France
dc.contributor.editoruniversityotherComputer Vision Laboratory - ETHZ [Zurich];Suisse
dc.contributor.editoruniversityotherSignal Processing Lab (DSP) Wuhan University Luojia Mountain,;Chine
dc.description.abstractenWe introduce the hierarchical Markov aspect model (HMAM), a computationally efficient graphical model for densely labeling large remote sensing images with their underlying terrain classes. HMAM resolves local ambiguities efficiently by combining the benefits of quadtree representations and aspect models--the former incorporate multiscale visual features and hierarchical smoothing to provide improved local label consistency, while the latter sharpen the labelings by focusing them on the classes that are most relevant for the broader local image context. The full HMAM model takes a grid of local hierarchical Markov quadtrees over image patches and augments it by incorporating a probabilistic latent semantic analysis aspect model over a larger local image tile at each level of the quadtree forest. Bag-of-word visual features are extracted for each level and patch, and given these, the parent-child transition probabilities from the quadtree and the label probabilities from the tile-level aspect models, an efficient forwards-backwards inference pass allows local posteriors for the class labels to be obtained for each patch. Variational expectation-maximization is then used to train the complete model from either pixel-level or tile-keyword-level labelings. Experiments on a complete TerraSAR-X synthetic aperture radar terrain map with pixel-level ground truth show that HMAM is both accurate and efficient, providing significantly better results than comparable single-scale aspect models with only a modest increase in training and test complexity. Keyword-level training greatly reduces the cost of providing training data with little loss of accuracy relative to pixel-level training.en
dc.relation.isversionofjnlnameIEEE Transactions on Image Processing
dc.subject.ddclabelProbabilités et mathématiques appliquéesen

Files in this item


There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record