@techreport{TR1862162, title = "Visual Analysis of Volume Ensembles Based on Local Features", author = "Johanna Schmidt and Bernhard Fr\"{o}hler and Reinhold Preiner and Johannes Kehrer and Eduard Gr\"{o}ller and Stefan Bruckner and Christoph Heinzl", year = "2016", abstract = "Ensemble datasets describe a specific phenomenon (e.g., a simulation scenario or a measurements series) through a large set of individual ensemble members. These individual members typically do not differ too much from each other but rather feature slightly changing characteristics. In many cases, the ensemble members are defined in 3D space, which implies severe challenges when exploring the complete ensembles such as handling occlusions, focus and context or its sheer datasize. In this paper we address these challenges and put our focus on the exploration of local features in 3D volumetric ensemble datasets, not only by visualizing local characteristics, but also by identifying connections to other local features with similar characteristics in the data. We evaluate the variance in the dataset and use the the spatial median (medoid) of the ensemble to visualize the differences in the dataset. This medoid is subsequently used as a representative of the ensemble in 3D. The variance information is used to guide users during the exploration, as regions of high variance also indicate larger changes within the ensemble members. The local characteristics of the regions can be explored by using our proposed 3D probing widgets. These widgets consist of a 3D sphere, which can be positioned at any point in 3D space. While moving a widget, the local data characteristics at the corresponding position are shown in a separate detail view, which depicts the local outliers and their surfaces in comparison to the medoid surface. The 3D probing widgets can also be fixed at a user-defined position of interest. The fixed probing widgets are arranged in a similarity graph to indicate similar local data characteristics. The similarity graph thus allows to explore whether high variances in a certain region are caused by the same dataset members or not. Finally, it is also possible to compare a single member against the rest of the ensemble. We evaluate our technique through two demonstration cases using volumetric multi-label segmentation mask datasets, two from the industrial domain and two from the medical domain.", month = may, number = "TR-186-2-16-2", address = "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria", institution = "Institute of Computer Graphics and Algorithms, Vienna University of Technology ", note = "human contact: technical-report@cg.tuwien.ac.at", keywords = "ensemble visualization, guided local exploration, variance analysis", URL = "https://www.cg.tuwien.ac.at/research/publications/2016/TR1862162/", } @article{beham-2014-cupid, title = "Cupid: Cluster-based Exploration of Geometry Generators with Parallel Coordinates and Radial Trees", author = "Michael Beham and Wolfgang Herzner and Eduard Gr\"{o}ller and Johannes Kehrer", year = "2014", abstract = "Geometry generators are commonly used in video games and evaluation systems for computer vision to create geometric shapes such as terrains, vegetation or airplanes. The parameters of the generator are often sampled automatically which can lead to many similar or unwanted geometric shapes. In this paper, we propose a novel visual exploration approach that combines the abstract parameter space of the geometry generator with the resulting 3D shapes in a composite visualization. Similar geometric shapes are first grouped using hierarchical clustering and then nested within an illustrative parallel coordinates visualization. This helps the user to study the sensitivity of the generator with respect to its parameter space and to identify invalid parameter settings. Starting from a compact overview representation, the user can iteratively drill-down into local shape differences by clicking on the respective clusters. Additionally, a linked radial tree gives an overview of the cluster hierarchy and enables the user to manually split or merge clusters. We evaluate our approach by exploring the parameter space of a cup generator and provide feedback from domain experts.", month = dec, journal = "IEEE Transactions on Visualization and Computer Graphics", volume = "20", number = "12", issn = "1077-2626", pages = "1693--1702 ", keywords = "3D shape analysis, radial trees, hierarchical clustering, illustrative parallel coordinates, composite visualization", URL = "https://www.cg.tuwien.ac.at/research/publications/2014/beham-2014-cupid/", } @talk{Kehrer-2014-CSD, title = "Interactive Visual Analysis of Complex Scientific Data", author = "Johannes Kehrer", year = "2014", event = "TU M\"{u}nchen", location = "Munich, Germany", URL = "https://www.cg.tuwien.ac.at/research/publications/2014/Kehrer-2014-CSD/", } @article{kehrer-2013-SBC, title = "A Model for Structure-based Comparison of Many Categories in Small-Multiple Displays", author = "Johannes Kehrer and Harald Piringer and Wolfgang Berger and Eduard Gr\"{o}ller", year = "2013", abstract = "Many application domains deal with multi-variate data that consists of both categorical and numerical information. Small-multiple displays are a powerful concept for comparing such data by juxtaposition. For comparison by overlay or by explicit encoding of computed differences, however, a specification of references is necessary. In this paper, we present a formal model for defining semantically meaningful comparisons between many categories in a small-multiple display. Based on pivotized data that are hierarchically partitioned by the categories assigned to the x and y axis of the display, we propose two alternatives for structure-based comparison within this hierarchy. With an absolute reference specification, categories are compared to a fixed reference category. With a relative reference specification, in contrast, a semantic ordering of the categories is considered when comparing them either to the previous or subsequent category each. Both reference specifications can be defined at multiple levels of the hierarchy (including aggregated summaries), enabling a multitude of useful comparisons. We demonstrate the general applicability of our model in several application examples using different visualizations that compare data by overlay or explicit encoding of differences.", month = dec, journal = "IEEE Transactions on Visualization and Computer Graphics", volume = "19", number = "12", pages = "2287--2296", URL = "https://www.cg.tuwien.ac.at/research/publications/2013/kehrer-2013-SBC/", } @WorkshopTalk{oeltze-2013-tut, title = "IEEE VIS Tutorial on Interactive Visual Analysis of Scientific Data", author = "Steffen Oeltze and Helwig Hauser and Johannes Kehrer", year = "2013", abstract = "In a growing number of application areas, a subject or phenomenon is investigated by means of multiple datasets being acquired over time (spatiotemporal), comprising several attributes per data point (multi-variate), stemming from different data sources (multi-modal) or multiple simulation runs (multi-run/ensemble). Interactive visual analysis (IVA) comprises concepts and techniques for a user-guided knowledge discovery in such complex data. Through a tight feedback loop of computation, visualization and user interaction, it provides new insight into the data and serves as a vehicle for hypotheses generation or validation. It is often implemented via a multiple coordinated view framework where each view is equipped with interactive drill-down operations for focusing on data features. Two classes of views are integrated: physical views, such as direct volume rendering, show information in the context of the spatiotemporal observation space while attribute views, such as scatter plots and parallel coordinates, show relationships between multiple data attributes. The user may drill-down the data by selecting interesting regions of the observation space or attribute ranges leading to a consistent highlighting of this selection in all other views (brushing-and-linking). Three patterns of explorative/analytical procedures may be accomplished by doing so. In a feature localization, the user searches for places in the 3D/4D observation space where certain attribute values are present. In a multi-variate analysis, relations between data attributes are investigated, e.g., by searching for correla- tions. In a local investigation, the user inspects the values of selected attributes with respect to certain spatiotemporal subsets of the observation space. In this tutorial, we discuss examples for successful applications of IVA to scientific data from various fields: climate research, medicine, epidemiology, and flow simulation / computation, in particular for automotive engineering. We base our discussions on a theoretical foundation of IVA which helps the tutorial attendees in transferring the subject matter to their own data and application area. In the course of the tutorial, the attendees will become acquainted with techniques from statistics and knowledge discovery, which proved to be particularly useful for a specific IVA application. The tutorial further comprises an overview of off-the-shelf IVA solutions, which may be be particularly interesting for visualization practitioners. It is concluded by a summary of the gained knowledge and a discussion of open problems in IVA of scientific data. The tutorial slides will be available at: http://tinyurl.com/SciDataIVA13", month = oct, event = "IEEE VisWeek", location = "Atlanta, Georgia, USA", URL = "https://www.cg.tuwien.ac.at/research/publications/2013/oeltze-2013-tut/", } @article{borgo-2013-gly, title = "Glyph-based Visualization: Foundations, Design Guidelines, Techniques and Applications", author = "Rita Borgo and Johannes Kehrer and David H.S. Chung and Eamonn Maguire and Robert S. Laramee and Helwig Hauser and Matthew Ward and Min Chen", year = "2013", abstract = "This state of the art report focuses on glyph-based visualization, a common form of visual design where a data set is depicted by a collection of visual objects referred to as glyphs. Its major strength is that patterns of multivariate data involving more than two attribute dimensions can often be more readily perceived in the context of a spatial relationship, whereas many techniques for spatial data such as direct volume rendering find difficult to depict with multivariate or multi-field data, and many techniques for non-spatial data such as parallel coordinates are less able to convey spatial relationships encoded in the data. This report fills several major gaps in the literature, drawing the link between the fundamental concepts in semiotics and the broad spectrum of glyph-based visualization, reviewing existing design guidelines and implementation techniques, and surveying the use of glyph-based visualization in many applications.", month = may, journal = "Eurographics State of the Art Reports", note = "http://diglib.eg.org/EG/DL/conf/EG2013/stars/039-063.pdf", publisher = "Eurographics Association", series = "EG STARs", pages = "39--63", URL = "https://www.cg.tuwien.ac.at/research/publications/2013/borgo-2013-gly/", } @article{Kehrer-2013-STAR, title = "Visualization and Visual Analysis of Multi-faceted Scientific Data: A Survey", author = "Johannes Kehrer and Helwig Hauser", year = "2013", abstract = "Visualization and visual analysis play important roles in exploring, analyzing and presenting scientific data. In many disciplines, data and model scenarios are becoming multi-faceted: data are often spatio-temporal and multi-variate; they stem from different data sources (multi-modal data), from multiple simulation runs (multi-run/ensemble data), or from multi-physics simulations of interacting phenomena (multi-model data resulting from coupled simulation models). Also, data can be of different dimensionality or structured on various types of grids that need to be related or fused in the visualization. This heterogeneity of data characteristics presents new opportunities as well as technical challenges for visualization research. Visualization and interaction techniques are thus often combined with computational analysis. In this survey, we study existing methods for visualization and interactive visual analysis of multi-faceted scientific data. Based on a thorough literature review, a categorization of approaches is proposed. We cover a wide range of fields and discuss to which degree the different challenges are matched with existing solutions for visualization and visual analysis. This leads to conclusions with respect to promising research directions, for instance, to pursue new solutions for multi-run and multi-model data as well as techniques that support a multitude of facets.", month = mar, issn = "1077-2626", journal = "IEEE Transactions on Visualization and Computer Graphics", note = "Spotlight paper of the March issue of TVCG", number = "3", volume = "19", pages = "495--513", URL = "https://www.cg.tuwien.ac.at/research/publications/2013/Kehrer-2013-STAR/", } @talk{kehrer-2013-IVA, title = "Visual Analysis of Multi-faceted Scientific Data: Challenges and Trends", author = "Johannes Kehrer", year = "2013", event = "Karlsruhe Institute of Technology", location = "Karlsruhe, Germany", URL = "https://www.cg.tuwien.ac.at/research/publications/2013/kehrer-2013-IVA/", } @article{kehrer-2011-vai, title = "Interactive Visual Analysis of Heterogeneous Scientific Data across an Interface", author = "Johannes Kehrer and Philipp Muigg and Helmut Doleisch and Helwig Hauser", year = "2011", abstract = "In this paper, we present a systematic approach to the interactive visual exploration and analysis of heterogeneous scientific data. Based on a setup of coordinated multiple views (with linking and brushing) and heterogeneous data which consists of two blocks of scientific data (e.g., 2D and 3D data), we enable the joint, feature-based investigation across an interface. The interface specifies (a) which items in the one part of the data are related to which items in the other part, and vice versa, (b) how selections (in terms of feature extraction) are transferred between the two parts of the data, and (c) how interaction is realized during the visual analysis. We also propose strategies for visual analysis across an interface resulting in interactive and iterative refinement of features specified in different parts of the data. We demonstrate the usefulness of our approach in the context of two visual analysis scenarios with heterogeneous scientific data, i.e., a multi-run climate simulation and a complex simulation of fluid-structure interaction.", month = jul, journal = "IEEE Transaction on Visualization and Computer Graphics", volume = "17", number = "7", issn = "1077-2626", pages = "934--946", URL = "https://www.cg.tuwien.ac.at/research/publications/2011/kehrer-2011-vai/", } @mastersthesis{Kehrer-2007-Mas, title = "Integrating Interactive Visual Analysis of Large Time Series Data into the SimVis System", author = "Johannes Kehrer", year = "2007", abstract = "Massive amounts of complex time-dependent information arise in various areas of business, science and engineering. These time series datasets commonly result from the measurement, modeling or the simulation of dynamic processes and contain multiple attributes changing over time. Examples are meteorological data, climate data, financial data, census data, or medical data, to name a few. In this thesis the CurveView for the enhanced interactive visual analysis of multidimensional and large time series data is presented. Two approaches are proposed, one for the interactive visual representation of the data, and so-called brushing techniques allowing the user to select certain interesting subsets of the data (i.e., features) in an intuitive and interactive way. The goals are to enable analysts to gain insight into their data sets, to create, verify or reject hypotheses based on the data, and to explore the temporal evolution of different attributes in order to detect expected structures and to discover unexpected features. The presented solution is integrated into SimVis, a multiple-views system for the visual analysis of time-dependent simulation results. The data is visualized using focus+context visualization techniques: important or selected portions of the data (focus) are visually accented, while the rest of the data (context) is shown in a less prominent style. In doing so, enhanced navigation and orientation is provided to the user. By the application of customizable transfer functions, general data trends, visual structures and patterns can be emphasized even within dense regions of the visualization. On the other hand, so-called outliers, which denote time series in low populated areas of the display or important (ie, brushed) data items hidden in regions of context information, are discriminable in the visualization. By the application of binning techniques large amounts of time-dependent information are transformed into a reduced but still meaningful representation which can be depicted at interactive frame rates. Furthermore, interactive brushing techniques are provided to the user for analysis purposes. Thus, complex time-dependent features can be specified by applying fuzzy classification to the time series data. Two kinds of brushes exist in the CurveView: similarity-based brushes where time series are classified according to their similarity to a user-defined pattern directly sketched in the view; and time step brushes, which select time series running through a certain area of the view. In SimVis, the interrelations between the specified features in multiple time-dependent dimensions can be analyzed visually using multiple linked views that show different attributes (i.e., dimensions) of the data.", month = oct, address = "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria", school = "Institute of Computer Graphics and Algorithms, Vienna University of Technology ", URL = "https://www.cg.tuwien.ac.at/research/publications/2007/Kehrer-2007-Mas/", }