GI-Forum SS 2016

Date	Topic	Presenter	Institution
12.04.2016	Institute meeting
19.04.2016	Anonymous Publish-Subscribe	Prof. Mathias Fischer	Universität Münster
	Publish-subscribe is an increasingly popular messaging pattern for distributed systems, supporting scalable and extensible programming, and optimal spatial, temporal, and control-flow decoupling of distributed components. Publish-subscribe middleware and methods were extended towards supporting security, in particular confidentiality, and increased availability, yet few prior works addressed anonymity of participants. Anonymity of senders and receivers may however be crucial, e.g., for supporting freedom of expression in regimes where political repression and censorship prevail. In this presentation, we review basic security and privacy requirements and introduce a new attacker model based on statistical disclosure, used to challenge anonymity. We elaborate on design options for privacy-preserving publish-subscribe systems and present a novel system that leverages peer-to-peer networking concepts; this novel approach protects subscriber anonymity by means of Probabilistic Forwarding (PF) and through a novel so-called Shell Game (SG) algorithm. We verify our solution against the requirements and provide a simulation-based analysis of the effectiveness of our approaches in light of our attacker model. The results show that the SG algorithm efficiently protects subscriber anonymity and that anonymity sets can be adjusted via PF.
26.04.2016	Gesture Analysis for Human Computer Interaction and Humanities Research	Priv-Doz. Dr. Oliver Schreer	Fraunhofer HHI
	Gesture analysis is a very popular research domain with applications in many different areas. The talk is divided in two parts representing the major research activities of Immersive Media & Communications Group at Vision & Imaging Technologies Department at Fraunhofer Heinrich Hertz Institute. Thanks to the availability of consumer depth sensors, recognition and tracking of humans and gestural behaviour entered many fields of applications such as entertainment, education, medical, automotive or industry 4.0. By exploiting depth and video information of commercial sensors, quite robust and complex recognition tasks can be solved. In the first part of the talk, an overview of concepts and approaches for touchless interaction in different fields of applications is given. The analysis of gestural behaviour plays an important role in humanities research, such as psycho-linguistic or psycho-therapy. In these communities, terra bytes of video data are available that have to be annotated and evaluated for various research questions. The annotation of gestural behaviour is a very time consuming task and therefore, automatic video analysis can speed-up the annotation process and avoid time consuming manual annotation of hours of video. In the second part of the talk, a frame work for automatic video analysis of single video streams is presented that aims at supporting humanities researcher to reduce annotation time.
03.05.2016	No GI-Forum
10.05.2016	UAV - paradigm shift in geodata acquisition	Prof. Rohan Benett, Prof. Markus Gerke	University of Twente
	The last decade saw tremendous developments regarding sensor technology and platforms to be used for earth observation. This talk will focus on very high resolution data acquisition and processing - down to the centimetre as observed from unmanned aerial vehicles (UAV). For many application fields within the geospatial domain, the use of UAVs opens the door for enhanced or even new methodologies. Those platforms are helping us to close gaps in the observation of the (man-made) environment in many aspects: they are flexible regarding the realisation of different viewing angles (perspectives), and can provide observations at very high spatial and temporal resolution. We will show examples from automated building damage mapping, informal settlement monitoring, or mapping for land administration.
17.05.2016	Pentecost Holidays
24.05.2016	Institute meeting
31.05.2016	EarthDB: Building a "Digital Earth" for Science	Prof. James Frew	University of California
	EarthDB is a prototype for a SciDB database of primary Earth observation data. Primary data is original sensor outputs or human observations, not subject to any reformatting, reprojection, aggregation, or other transformations beyond those required to digitize and ingest them. EarthDB should allow common Earth science analytical operations (e.g., coordinate transformations, aggregation, spatial algebra, etc.) or data fusion (e.g., joins across multiple heterogeneous data types and representations) to be expressed as database queries against original observations, providing heretofore unavailable flexibility and traceability. This talk describes progress towards and challenges to achieving this goal.
07.06.2016	Visualizing Distance of Off-screen Landmarks on Mobile Devices: Visual Variables and Their Effects	Prof. Rui Li	University at Albany
	Navigation systems on mobile devices are becoming widely popular in both academic studies and industrial development. Designs attempting to display off-screen landmarks on screen as a way to overcome the limit of the small display have emerged. Some of these designs not only decrease the frequency of zooming but also contribute to users’ sense of direction during navigation by showing the direction to off-screen objects. Built upon previous work, the following work focuses on improved designs that not only display the direction but also the concept of distance of off-screen landmarks. This series of studies investigate visual variables, fundamental cartographic elements on symbol design and visualization, on perception of the distance. These employed visual variables include size, shape, saturation, transparency, and vagueness in our studies. These off-screen landmarks change in values of each visual variable as an indicator of the distance to user’s mapped location. In order to compare the efficiency of visualizing the distance concept, we first design two types of mechanism: one is using ratio level of measurement to create off-screen landmark symbols that the value of a specific visual variable changes proportionally according to their distances; the other type is using ordinal level of measurement which only uses three different values in a visual variable to represent three different ranges of distance. User studies are then carried out first to evaluate the influence of these two levels of measurement, and then to investigate each visual variable’s effects on perceiving the distance. The goal of these studies can suggest effective design of off-screen landmarks on mobile devices contributing to the ongoing WayTO project. A further step of this study is the integration of the suggested design with routing services on mobile devices to support pedestrian wayfinding through orientation.
14.06.2016	Closing the Loop – How to model the subsurface	Dr. Tobias Rudolf	EON Gas Storage
	Dr. Rudolf will present the subsurface modelling workflow and give examples of applying the software Petrel in three different examples. These examples will highlight the challenges EON faces in the exploration and production as well as in the storage industry.
21.06.2016	Effects of visual detail in topographic maps on spatial cognition and memory	Prof. Dr. Frank Dickmann Prof. Dr. Lars Kuchinke	Ruhr Universität Bochum
	Orientation in and navigation through an environment takes often place on the basis of previous experiences with this environment. The use of topographic maps for example, offers a map reader a huge amount of spatial information to build a mental representation of an unexplored environment. After perception and filtering, this information is embodied in so-called cognitive maps. A cognitive map represents information about spatial objects, the relations and distances between them. The elements of a cognitive map are structured and can be mentally inspected. It is well documented that cognitive maps based on information learned from maps enclose spatial distortions. A specific map information can thus either be advantageous or disadvantageous in form of stored distortion errors. A number of studies were conducted in this interdisciplinary project between the geography and cognitive science. With the use of different memory paradigms the effect of visual detail in topographic maps (such as topographic detail and grid lines as a structuring map elements) on the memory performance of map users was examined. Results clearly point to a modulation of spatial memory by manipulated visual detail in these maps. Moreover, next to the analyses of recall and recognition of spatial information, eye-tracking is used to examine how map users process this map information. Eye-tracking data show that topographic detail as well as grids trigger attentional shifts towards specific (to-be-learned) map content. In addition, measures like the number of fixations, time to first fixation and fixation duration reveal that spatial information from maps modulates the fixation pattern and thus the way of information is encoded and processed from topographic maps. These differences in the fixation pattern and the attentional focus seem to be pivotal to increase the spatial accuracy of a cognitive map.
28.06.2016	No GI-Forum
05.07.2016	Perceptual-motor Expertise and Sports Performance	Prof. Karen Zentrgraf	Universität Münster
	In Summer 2014, about 35 million people in Germany watched on TV as the German football team became world champion for the fourth time. What are the differences between the people playing and the people watching?In this presentation, we will investigate answers to this question on several levels. We will review recent research based on Ericsson’s approach from cognitive psychology on the domain specificity of expertise, and we will also discuss the importance of ‘deliberate practice’. Findings in this area show that experts in specific sports disciplines exhibit better performance and trainability in cognitive tests (spatial attention, inhibition, etc.) that are fairly independent of individual sports disciplines. In addition, we will present results from neuroscience that link sports-related perceptual-cognitive performance with underlying central nervous processes and that emphasise the role of premotor-parietal-cerebral networks. We will also discuss whether exceptional performance performance particularly in play sports disciplines are a result of athletes being ‘multitasker’, ‘automiser’ and/or ‘fast processors’ in motor and cognitive demands. A valid model of expertise in sports also has to consider exceptional performances as optimal adaptation of the entire system to the specific requirements of the environment.
12.07.2016	Land cover monitoring using satellite image time series	Dr. Victor Maus	National Institute for Space Research (INPE), Brazil e-sensing
	Land system change has a wide range of impacts on Earth system components. Tropical forests in particular are crucial ecosystems for climate regulation, global biodiversity, and hydrological cycling. The current large-scale classification systems such as MODIS Land Cover and GLC 2000 have limitations and their accuracy is not sufficient for land cover change analysis. Therefore, new techniques for improving land cover products are urgently needed. Here I present some contributions on Big Earth Observation (EO) Data handling, new methods for satellite image time series analysis, and our most recent founds about the land cover dynamics in the Amazon Biome at Mato Grosso, Brazil.
19.07.2016	Institute meeting