Fraunhofer-Institut für Optronik, Systemtechnik und Bildauswertung IOSB

Filters

Reset filters

Settings
Now showing 1 - 10 of 16
  • Publication
    Extending StructureNet to generate physically feasible 3D shapes
    ( 2021)
    Koch, Jannik
    ;
    Haraké, Laura
    ;
    Jung, Alisa
    ;
    Dachsbacher, Carsten
    StructureNet is a recently introduced n-ary graph network that generates 3D structures with awareness of geometric part relationships and promotes reasonable interactions between shape parts. However, depending on the inferred latent space, the generated objects may lack physical feasibility, since parts might be detached or not arranged in a load-bearing manner. We extend StructureNet's training method to optimize the physical feasibility of these shapes by adapting its loss function to measure the structural intactness. Two new changes are hereby introduced and applied on disjunctive shape parts: First, for the physical feasibility of linked parts, forces acting between them are determined. Considering static equilibrium, compression and friction, they are assembled in a constraint system as the Measure of Infeasibility. The required interfaces between these parts are identified using Constructive Solid Geometry. Secondly, we define a novel metric called Hover Penalty that detects and penalizes unconnected shape parts to improve the overall feasibility. The extended StructureNet is trained on PartNet's chair data set, using a bounding box representation for the geometry. We demonstrate first results that indicate a significant reduction of hovering shape parts and a promising correction of shapes that would be physically infeasible.
  • Publication
    Automatic Generation of Training Data for Land Use and Land Cover Classification by Fusing Heterogeneous Data Sets
    ( 2020)
    Schmitz, Sylvia
    ;
    Weinmann, Martin
    ;
    Weidner, Uwe
    ;
    Hammer, Horst
    ;
    Thiele, Antje
    Nowadays, automatic classification of remote sensing data can efficiently produce maps of land use and land cover, which provide an essential source of information in the field of environmental sciences. Most state-of-the-art algorithms use supervised learning methods that require a large amount of annotated training data. In order to avoid time-consuming manual labelling, we propose a method for the automatic annotation of remote sensing data that relies on available land use and land cover information. Using the example of automatic labelling of SAR data, we show how the Dempster-Shafer evidence theory can be used to fuse information from different land use and land cover products into one training data set. Our results confirm that the combination of information from OpenStreetMap, CORINE Land Cover 2018, Global Surface Water and the SAR data itself leads to reliable class assignments, and that this combination outperforms each considered single land use and land cover product.
  • Publication
    From multi-sensor aerial data to thermal and infrared simulation of semantic 3D models: Towards identification of urban heat islands
    ( 2020)
    Bulatov, Dimitri
    ;
    Burkard, Eva
    ;
    Ilehag, Rebecca
    ;
    Kottler, Benedikt
    ;
    Helmholz, Petra
    Urban heat islands degrade the quality of life in many urban centers. To achieve their detection in urban canopy and to predict their development in the future, infrared simulation turns out to be a suitable tool. For simulation of the temperature, various scene properties must be taken into account. Starting at raw sensor data acquired from the air, we developed an end-to-end pipeline to the semantic mesh, in which temperatures and radiance can be simulated depending on actual weather data and initial conditions and which has a potential to track the urban heat islands. To acquire the mesh, we focus on retrieving land cover classes and 3D geometry. The land cover map helps to identify buildings, to update the existing geographic maps, and to analyze building roofs with respect to their materials and thus, sustainability. The 3D geometry basically presupposes storing the scene efficiently into triangles. For each triangle, we are not only interested in material properties, but also in neighborhood relations allowing to model heat conduction. Together with terms for convection and radiation, we formulate the heat balance equation and compute the surface temperature as a function of time. The pipeline was tested on a dataset from a large Australian city exhibiting most properties which bear risks to contribute to heat islands: Its location in a subtropical (Mediterranean) climate zone, rapidly growing population, and, at least initially, a certain lack of sensibility towards sustainable management of resources and materials. To analyze both latter factors, two intermediate results from our method, namely tracking urbanization degree and identification of common roofing materials, are addressed and thoroughly evaluated in the dataset. It could be deduced that the area occupied by buildings increased by roughly 5% and that roughly every 6th building has a steel roof. Finally, high similarities with the ground truth were achieved both for temperature curves in some 20 test points and for large-scale evaluation. Deviations from the ground truth emerge in case of building roofs leading to the conclusion that the inner model assumption could be less accurate and, therefore, runs the danger to increase the urban heat island effect.
  • Publication
    Selbstkalibrierung mobiler Multisensorsysteme mittelsgeometrischer 3D-Merkmale
    (KIT, 2020)
    Hillemann, Markus
    Ein mobiles Multisensorsystem ermöglicht die effiziente, räumliche Erfassung von Objekten und der Umgebung. Die Kalibrierung des mobilen Multisensorsystems ist ein notwendiger Vorverarbeitungsschritt für die Sensordatenfusion und für genaue räumliche Erfassungen. Bei herkömmlichen Verfahren kalibrieren Experten das mobile Multisensorsystem in aufwändigen Prozeduren vor Verwendung durch Aufnahmen eines Kalibrierobjektes mit bekannter Form. Im Gegensatz zu solchen objektbasierten Kalibrierungen ist eine Selbstkalibrierung praktikabler, zeitsparender und bestimmt die gesuchten Parameter mit höherer Aktualität.
  • Publication
    High-pulse-energy actively Q-switched polarization-maintaining Tm3+-doped silica fiber laser
    ( 2020)
    Schneider, Julian
    ;
    Forster, Patrick
    ;
    Eichhorn, Marc
    ;
    Kieleck, Christelle
    A diode-pumped Q-switched thulium-doped fiber laser is reported providing 43 ns short pulses with pulse energies of 800 µJ at a wavelength of 2050 nm.
  • Publication
    Superpoints in RANSAC planes: A new approach for ground surface extraction exemplified on point classification and context-aware reconstruction
    ( 2020)
    Bulatov, Dimitri
    ;
    Stütz, Dominik
    ;
    Lucks, Lukas
    ;
    Weinmann, Martin
    In point clouds obtained from airborne data, the ground points have traditionally been identified as local minima of the altitude. Subsequently, the 2.5D digital terrain models have been computed by approximation of a smooth surfaces from the ground points. But how can we handle purely 3D surfaces of cultural heritage monuments covered by vegetation or Alpine overhangs, where trees are not necessarily growing in bottom-to-top direction? We suggest a new approach based on a combination of superpoints and RANSAC implemented as a filtering procedure, which allows efficient handling of large, challenging point clouds without necessity of training data. If training data is available, covariance-based features, point histogram features, and dataset-dependent features as well as combinations thereof are applied to classify points. Results achieved with a Random Forest classifier and non-local optimization using Markov Random Fields are analyzed for two challenging datasets: an airborne laser scan and a photogrammetrically reconstructed point cloud. As an application, surface reconstruction from the thus cleaned point sets is demonstrated.
  • Publication
    Simultaneous identification of wind turbine vibrations by using seismic data, elastic modeling and laser Doppler vibrometry
    ( 2020)
    Zieger, Toni
    ;
    Nagel, S.
    ;
    Lutzmann, Peter
    ;
    Kaufmann, Ilja
    ;
    Ritter, J.
    ;
    Ummenhofer, T.
    ;
    Knödel, P.
    ;
    Fischer, Peter
    This work compares continuous seismic ground motion recordings over several months on top of the foundation and in the near field of a wind turbine (WT) at Pfinztal, Germany, with numerical tower vibration simulations and simultaneous optical measurements. We are able to distinguish between the excitation of eigenfrequencies of the tower-nacelle system and the influence of the blade rotation on seismic data by analyzing different wind and turbine conditions. We can allocate most of the major spectral peaks to either different bending modes of the tower, flapwise, and edgewise bending modes of the blades or multiples of the blade-passing frequency after comparing seismic recordings with tower simulation models. These simulations of dynamic properties of the tower are based on linear modal analysis performed with finite beam elements. To validate our interpretations of the comparison of seismic recordings and simulations, we use optical measurements of a laser Doppler vibrometer at the tower of the turbine at a height of about 20 m. The calculated power spectrum of the tower vibrations confirms our interpretation of the seismic peaks regarding the tower bending modes. This work gives a new understanding of the source mechanisms of WT-induced ground motions and their influence on seismic data by using an interdisciplinary approach. Thus, our results may be used for structural health purposes as well as the development of structural damping methods, which can also reduce ground motion emissions from WTs. Furthermore, it demonstrates how numerical simulations of wind turbines can be validated by using seismic recordings and laser Doppler vibrometry.
  • Publication
    High-pulse-energy Q-switched Ho3+:YAG laser
    ( 2020)
    Büker, H.
    ;
    Braesicke, P.
    ;
    Kieleck, Christelle
    ;
    Eichhorn, Marc
    We report on a high-energy Ho3+:YAG laser with a maximum average power of 20.1W at a central wavelength of 2.09µm and M2 < 1.5. In Q-switched operation a maximum pulse energy of 15mJ at 700Hz repetition rate was achieved.
  • Publication
    Incorporating interferometric coherence into LULC classification of airborne PolSAR-images using fully convolutional networks
    ( 2020)
    Schmitz, Sylvia
    ;
    Weinmann, Martin
    ;
    Thiele, Antje
    Inspired by the application of state-of-the-art Fully Convolutional Networks (FCNs) for the semantic segmentation of high-resolution optical imagery, recent works transfer this methodology successfully to pixel-wise land use and land cover (LULC) classification of PolSAR data. So far, mainly single PolSAR images are included in the FCN-based classification processes. To further increase classification accuracy, this paper presents an approach for integrating interferometric coherence derived from co-registered image pairs into a FCN-based classification framework. A network based on an encoder-decoder structure with two separated encoder branches is presented for this task. It extracts features from polarimetric backscattering intensities on the one hand and interferometric coherence on the other hand. Based on a joint representation of the complementary features pixel-wise classification is performed. To overcome the scarcity of labelled SAR data for training and testing, annotations are generated automatically by fusing available LULC products. Experimental evaluation is performed on high-resolution airborne SAR data, captured over the German Wadden Sea. The results demonstrate that the proposed model produces smooth and accurate classification maps. A comparison with a single-branch FCN model indicates that the appropriate integration of interferometric coherence enables the improvement of classification performance.
  • Publication
    Strategic optimization of convolutional neural networks for hyperspectral land cover classification
    ( 2020)
    Bühler, C.
    ;
    Schenkel, F.
    ;
    Groß, W.
    ;
    Schaab, G.
    ;
    Middelmann, W.
    Hyperspectral data recorded by future earth observation satellites will have up to hundreds of narrow bands that cover a wide range of the electromagnetic spectrum. The spatial resolution (around 30 meters) of such data, however, can impede the integration of the spatial domain for a classification due to spectrally mixed pixels and blurred edges in the data. Hence, the ability of performing a meaningful classification only relying on spectral information is important. In this study, a model for the spectral classification of hyperspectral data is derived by strategically optimizing a convolutional neural network (1D-CNN). The model is pre-trained and optimized on imagery of different nuts, beans, peas and dried fruits recorded with the Cubert ButterflEye X2 sensor. Subsequently, airborne hyperspectral datasets (Greding, Indian Pines and Pavia University) are used to evaluate the CNN's capability of transfer learning. For that, the datasets are classified with the pre-trained weights and, for comparison, with the same model architecture but trained from scratch with random weights. The results show substantial differences in classification accuracies (from 71.8% to 99.8% overall accuracy) throughout the used datasets, mainly caused by variations in the number of training samples, the spectral separability of the classes as well as the existence of mixed pixels for one dataset. For the dataset that is classified least accurately, the greatest improvement with pre-training is achieved (difference of 3.3% in overall accuracy compared to the non-pre-trained model). For the dataset that is classified with the highest accuracy, no significant transfer learning was observed.