Centre de documentation
scientifique

Dépouillements

Detection of fallen trees in ALS point clouds using a Normalized Cut approach trained by simulation / Przemyslaw Polewski in ISPRS Journal of photogrammetry and remote sensing, vol 105 (July 2015)  

Public [article]  inISPRS Journal of photogrammetry and remote sensing > vol 105 (July 2015) . - pp 252 - 271 Titre : Detection of fallen trees in ALS point clouds using a Normalized Cut approach trained by simulation Type de document : Article/Communication Auteurs : Przemyslaw Polewski, Auteur ; Wei Yao, Auteur ; Marco Heurich, Auteur ; Peter Krzystek, Auteur ; Uwe Stilla, Auteur Année de publication : 2015 Article en page(s) : pp 252 - 271 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Lasergrammétrie [Termes IGN] arbre mort [Termes IGN] Bavière (Allemagne) [Termes IGN] détection automatique [Termes IGN] données lidar [Termes IGN] données localisées 3D [Termes IGN] inventaire forestier local [Termes IGN] parc naturel national [Termes IGN] semis de points Résumé : (auteur) Downed dead wood is regarded as an important part of forest ecosystems from an ecological perspective, which drives the need for investigating its spatial distribution. Based on several studies, Airborne Laser Scanning (ALS) has proven to be a valuable remote sensing technique for obtaining such information. This paper describes a unified approach to the detection of fallen trees from ALS point clouds based on merging short segments into whole stems using the Normalized Cut algorithm. We introduce a new method of defining the segment similarity function for the clustering procedure, where the attribute weights are learned from labeled data. Based on a relationship between Normalized Cut’s similarity function and a class of regression models, we show how to learn the similarity function by training a classifier. Furthermore, we propose using an appearance-based stopping criterion for the graph cut algorithm as an alternative to the standard Normalized Cut threshold approach. We set up a virtual fallen tree generation scheme to simulate complex forest scenarios with multiple overlapping fallen stems. This simulated data is then used as a basis to learn both the similarity function and the stopping criterion for Normalized Cut. We evaluate our approach on 5 plots from the strictly protected mixed mountain forest within the Bavarian Forest National Park using reference data obtained via a manual field inventory. The experimental results show that our method is able to detect up to 90% of fallen stems in plots having 30–40% overstory cover with a correctness exceeding 80%, even in quite complex forest scenes. Moreover, the performance for feature weights trained on simulated data is competitive with the case when the weights are calculated using a grid search on the test data, which indicates that the learned similarity function and stopping criterion can generalize well on new plots. Numéro de notice : A2015-703 Affiliation des auteurs : non IGN Thématique : FORET/IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1016/j.isprsjprs.2015.01.010 En ligne : https://doi.org/10.1016/j.isprsjprs.2015.01.010 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=78339 [article]

Semantic point cloud interpretation based on optimal neighborhoods, relevant features and efficient classifiers / Martin Weinmann in ISPRS Journal of photogrammetry and remote sensing, vol 105 (July 2015)  

Public [article]  inISPRS Journal of photogrammetry and remote sensing > vol 105 (July 2015) . - pp 286 - 304 Titre : Semantic point cloud interpretation based on optimal neighborhoods, relevant features and efficient classifiers Type de document : Article/Communication Auteurs : Martin Weinmann, Auteur ; Stefan Hinz, Auteur ; Clément Mallet , Auteur Année de publication : 2015 Article en page(s) : pp 286 - 304 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] classification barycentrique [Termes IGN] compréhension de l'image [Termes IGN] données localisées 3D [Termes IGN] environnement de développement [Termes IGN] extraction de traits caractéristiques [Termes IGN] opérateur sémantique [Termes IGN] scène [Termes IGN] semis de points [Termes IGN] voisinage (relation topologique) Mots-clés libres : Point cloud  Neighborhood selection  Feature extraction  Feature selection  Classification  3D scene analysis Résumé : (auteur) 3D scene analysis in terms of automatically assigning 3D points a respective semantic label has become a topic of great importance in photogrammetry, remote sensing, computer vision and robotics. In this paper, we address the issue of how to increase the distinctiveness of geometric features and select the most relevant ones among these for 3D scene analysis. We present a new, fully automated and versatile framework composed of four components: (i) neighborhood selection, (ii) feature extraction, (iii) feature selection and (iv) classification. For each component, we consider a variety of approaches which allow applicability in terms of simplicity, efficiency and reproducibility, so that end-users can easily apply the different components and do not require expert knowledge in the respective domains. In a detailed evaluation involving 7 neighborhood definitions, 21 geometric features, 7 approaches for feature selection, 10 classifiers and 2 benchmark datasets, we demonstrate that the selection of optimal neighborhoods for individual 3D points significantly improves the results of 3D scene analysis. Additionally, we show that the selection of adequate feature subsets may even further increase the quality of the derived results while significantly reducing both processing time and memory consumption. Numéro de notice : A2015-704 Affiliation des auteurs : LASTIG MATIS+Ext (2012-2019) Thématique : IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1016/j.isprsjprs.2015.01.016 Date de publication en ligne : 27/02/2015 En ligne : https://doi.org/10.1016/j.isprsjprs.2015.01.016 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=78340 [article]

Multiclass feature learning for hyperspectral image classification: Sparse and hierarchical solutions / Devis Tuia in ISPRS Journal of photogrammetry and remote sensing, vol 105 (July 2015)  

Public [article]  inISPRS Journal of photogrammetry and remote sensing > vol 105 (July 2015) . - pp 272 - 285 Titre : Multiclass feature learning for hyperspectral image classification: Sparse and hierarchical solutions Type de document : Article/Communication Auteurs : Devis Tuia, Auteur ; Rémi Flamary, Auteur ; Nicolas Courty, Auteur Année de publication : 2015 Article en page(s) : pp 272 - 285 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] apprentissage profond [Termes IGN] extraction de traits caractéristiques [Termes IGN] image hyperspectrale [Termes IGN] régression logistique Résumé : (auteur) In this paper, we tackle the question of discovering an effective set of spatial filters to solve hyperspectral classification problems. Instead of fixing a priori the filters and their parameters using expert knowledge, we let the model find them within random draws in the (possibly infinite) space of possible filters. We define an active set feature learner that includes in the model only features that improve the classifier. To this end, we consider a fast and linear classifier, multiclass logistic classification, and show that with a good representation (the filters discovered), such a simple classifier can reach at least state of the art performances. We apply the proposed active set learner in four hyperspectral image classification problems, including agricultural and urban classification at different resolutions, as well as multimodal data. We also propose a hierarchical setting, which allows to generate more complex banks of features that can better describe the nonlinearities present in the data. Numéro de notice : A2015-705 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1016/j.isprsjprs.2015.01.006 En ligne : https://doi.org/10.1016/j.isprsjprs.2015.01.006 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=78341 [article]

Random Forest and Rotation Forest for fully polarized SAR image classification using polarimetric and spatial features / Peijun Du in ISPRS Journal of photogrammetry and remote sensing, vol 105 (July 2015)  

Public [article]  inISPRS Journal of photogrammetry and remote sensing > vol 105 (July 2015) . - pp 38 - 53 Titre : Random Forest and Rotation Forest for fully polarized SAR image classification using polarimetric and spatial features Type de document : Article/Communication Auteurs : Peijun Du, Auteur ; Alim Samat, Auteur ; Björn Waske, Auteur ; Sicong Liu, Auteur ; Zhenhong Li, Auteur Année de publication : 2015 Article en page(s) : pp 38 - 53 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image radar et applications [Termes IGN] classification par forêts d'arbres décisionnels [Termes IGN] données polarimétriques [Termes IGN] image Radarsat [Termes IGN] polarimétrie radar [Termes IGN] Rotation Forest classification [Termes IGN] texture d'image Résumé : (auteur) Fully Polarimetric Synthetic Aperture Radar (PolSAR) has the advantages of all-weather, day and night observation and high resolution capabilities. The collected data are usually sorted in Sinclair matrix, coherence or covariance matrices which are directly related to physical properties of natural media and backscattering mechanism. Additional information related to the nature of scattering medium can be exploited through polarimetric decomposition theorems. Accordingly, PolSAR image classification gains increasing attentions from remote sensing communities in recent years. However, the above polarimetric measurements or parameters cannot provide sufficient information for accurate PolSAR image classification in some scenarios, e.g. in complex urban areas where different scattering mediums may exhibit similar PolSAR response due to couples of unavoidable reasons. Inspired by the complementarity between spectral and spatial features bringing remarkable improvements in optical image classification, the complementary information between polarimetric and spatial features may also contribute to PolSAR image classification. Therefore, the roles of textural features such as contrast, dissimilarity, homogeneity and local range, morphological profiles (MPs) in PolSAR image classification are investigated using two advanced ensemble learning (EL) classifiers: Random Forest and Rotation Forest. Supervised Wishart classifier and support vector machines (SVMs) are used as benchmark classifiers for the evaluation and comparison purposes. Experimental results with three Radarsat-2 images in quad polarization mode indicate that classification accuracies could be significantly increased by integrating spatial and polarimetric features using ensemble learning strategies. Rotation Forest can get better accuracy than SVM and Random Forest, in the meantime, Random Forest is much faster than Rotation Forest. Numéro de notice : A2015-706 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1016/j.isprsjprs.2015.03.002 En ligne : https://doi.org/10.1016/j.isprsjprs.2015.03.002 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=78342 [article]

Savannah woody structure modelling and mapping using multi-frequency (X-, C- and L-band) Synthetic Aperture Radar data / Laven Naidoo in ISPRS Journal of photogrammetry and remote sensing, vol 105 (July 2015)  

Public [article]  inISPRS Journal of photogrammetry and remote sensing > vol 105 (July 2015) . - pp 234 - 250 Titre : Savannah woody structure modelling and mapping using multi-frequency (X-, C- and L-band) Synthetic Aperture Radar data Type de document : Article/Communication Auteurs : Laven Naidoo, Auteur ; Renaud Mathieu, Auteur ; Russell Main, Auteur ; Waldo Kleynhans, Auteur ; et al., Auteur Année de publication : 2015 Article en page(s) : pp 234 - 250 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image radar et applications [Termes IGN] Afrique du sud (état) [Termes IGN] bande C [Termes IGN] bande L [Termes IGN] bande X [Termes IGN] biomasse [Termes IGN] canopée [Termes IGN] classification par forêts d'arbres décisionnels [Termes IGN] données lidar [Termes IGN] données localisées 3D [Termes IGN] image ALOS-PALSAR [Termes IGN] image radar moirée [Termes IGN] image Radarsat [Termes IGN] image TerraSAR-X [Termes IGN] savane Résumé : (auteur) Structural parameters of the woody component in African savannahs provide estimates of carbon stocks that are vital to the understanding of fuelwood reserves, which is the primary source of energy for 90% of households in South Africa (80% in Sub-Saharan Africa) and are at risk of over utilisation. The woody component can be characterised by various quantifiable woody structural parameters, such as tree cover, tree height, above ground biomass (AGB) or canopy volume, each been useful for different purposes. In contrast to the limited spatial coverage of ground-based approaches, remote sensing has the ability to sense the high spatio-temporal variability of e.g. woody canopy height, cover and biomass, as well as species diversity and phenological status – a defining but challenging set of characteristics typical of African savannahs. Active remote sensing systems (e.g. Light Detection and Ranging – LiDAR; Synthetic Aperture Radar – SAR), on the other hand, may be more effective in quantifying the savannah woody component because of their ability to sense within-canopy properties of the vegetation and its insensitivity to atmosphere and clouds and shadows. Additionally, the various components of a particular target’s structure can be sensed differently with SAR depending on the frequency or wavelength of the sensor being utilised. This study sought to test and compare the accuracy of modelling, in a Random Forest machine learning environment, woody above ground biomass (AGB), canopy cover (CC) and total canopy volume (TCV) in South African savannahs using a combination of X-band (TerraSAR-X), C-band (RADARSAT-2) and L-band (ALOS PALSAR) radar datasets. Training and validation data were derived from airborne LiDAR data to evaluate the SAR modelling accuracies. It was concluded that the L-band SAR frequency was more effective in the modelling of the CC (coefficient of determination or R2 of 0.77), TCV (R2 of 0.79) and AGB (R2 of 0.78) metrics in Southern African savannahs than the shorter wavelengths (X- and C-band) both as individual and combined (X + C-band) datasets. The addition of the shortest wavelengths also did not assist in the overall reduction of prediction error across different vegetation conditions (e.g. dense forested conditions, the dense shrubby layer and sparsely vegetated conditions). Although the integration of all three frequencies (X + C + L-band) yielded the best overall results for all three metrics (R2 = 0.83 for CC and AGB and R2 = 0.85 for TCV), the improvements were noticeable but marginal in comparison to the L-band alone. The results, thus, do not warrant the acquisition of all three SAR frequency datasets for tree structure monitoring in this environment. Numéro de notice : A2015-713 Affiliation des auteurs : non IGN Thématique : FORET/IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1016/j.isprsjprs.2015.04.007 En ligne : https://doi.org/10.1016/j.isprsjprs.2015.04.007 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=78353 [article]